USF Libraries
USF Digital Collections

Priming expectancies :

MISSING IMAGE

Material Information

Title:
Priming expectancies : effects on neurophysiological indices of expectancy violations and drinking behavior
Physical Description:
Book
Language:
English
Creator:
Brumback, Tyler
Publisher:
University of South Florida
Place of Publication:
Tampa, Fla
Publication Date:

Subjects

Subjects / Keywords:
Alcohol expectancies
Event-related potentials
P300
Semantic processing
PCA
Dissertations, Academic -- Psychology -- Masters -- USF   ( lcsh )
Genre:
non-fiction   ( marcgt )

Notes

Abstract:
ABSTRACT: Investigations of the anticipated effects of alcohol indicate that cognitive frameworks are highly correlated with drinking and other variables associated with alcohol use, explaining up to 50% of the variance in drinking outcomes (Goldman, Darkes, & Del Boca, 1999; Goldman, 2002; Goldman et al., 2006; Goldman, Reich, & Darkes, 2006). Furthermore, alcohol expectancies appear to mediate the relationship between a variety of risk factors, such as sensation seeking, and alcohol outcomes (Darkes, Greenbaum, & Goldman, 2004). The current study examined the relationship of these cognitive networks with a physiological index of expectancy violation Participants were presented with statements reflecting a wide range of alcohol outcome effects, which either violated or confirmed the participant's own set of alcohol expectancies, while the ERPs evoked by these stimuli were recorded. As predicted, the P300 amplitude elicited by negative alcohol expectancy stimuli was positively correlated with the degree of endorsement of positive/arousing expectancies on the self-report measure. That is, the higher the individual's positive/arousing expectancies, the larger the P300 elicited by stimuli asserting the negative effects of alcohol. There was no significant correlation, however, between P300 amplitude elicited by positive alcohol expectancy vi stimuli and the degree of endorsement of negative/sedating expectancies on the selfreport measure. In addition, individual differences relating to alcohol expectancies were examined as well. These results were able to identify specific stimuli that violated expectancies for each individual, as well as those that tended to violate expectancies in systematic ways across subjects. These findings provide a way forward for more precise assessment and prediction based on the well developed cognitive model of Alcohol Expectancies. In sum, variations in the amplitude of the P300 were consistent with the model of Alcohol Expectancies. Words imputing negative/sedating effects of alcohol elicited a large P300 in individuals with higher positive alcohol expectancies. By indexing the brain's electrophysiological response sensitive to expectancy violations, these findings demonstrate concordance between verbal measures of alcohol expectancies, which by their very nature are introspective, and a psychophysiological index of expectancy thought to operate automatically and to be independent of overt responding.
Thesis:
Thesis (M.A.)--University of South Florida, 2010.
Bibliography:
Includes bibliographical references.
System Details:
Mode of access: World Wide Web.
System Details:
System requirements: World Wide Web browser and PDF reader.
Statement of Responsibility:
by Tyler Brumback.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains X pages.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
usfldc doi - E14-SFE0003468
usfldc handle - e14.3468
System ID:
SFS0027783:00001


This item is only available as the following downloads:


Full Text
xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam 22 Ka 4500
controlfield tag 007 cr-bnu---uuuuu
008 s2010 flu s 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0003468
035
(OCoLC)
040
FHM
c FHM
049
FHMM
090
XX9999 (Online)
1 100
Brumback, Tyler.
0 245
Priming expectancies :
b effects on neurophysiological indices of expectancy violations and drinking behavior
h [electronic resource] /
by Tyler Brumback.
260
[Tampa, Fla] :
University of South Florida,
2010.
500
Title from PDF of title page.
Document formatted into pages; contains X pages.
502
Thesis (M.A.)--University of South Florida, 2010.
504
Includes bibliographical references.
516
Text (Electronic thesis) in PDF format.
538
Mode of access: World Wide Web.
System requirements: World Wide Web browser and PDF reader.
3 520
ABSTRACT: Investigations of the anticipated effects of alcohol indicate that cognitive frameworks are highly correlated with drinking and other variables associated with alcohol use, explaining up to 50% of the variance in drinking outcomes (Goldman, Darkes, & Del Boca, 1999; Goldman, 2002; Goldman et al., 2006; Goldman, Reich, & Darkes, 2006). Furthermore, alcohol expectancies appear to mediate the relationship between a variety of risk factors, such as sensation seeking, and alcohol outcomes (Darkes, Greenbaum, & Goldman, 2004). The current study examined the relationship of these cognitive networks with a physiological index of expectancy violation Participants were presented with statements reflecting a wide range of alcohol outcome effects, which either violated or confirmed the participant's own set of alcohol expectancies, while the ERPs evoked by these stimuli were recorded. As predicted, the P300 amplitude elicited by negative alcohol expectancy stimuli was positively correlated with the degree of endorsement of positive/arousing expectancies on the self-report measure. That is, the higher the individual's positive/arousing expectancies, the larger the P300 elicited by stimuli asserting the negative effects of alcohol. There was no significant correlation, however, between P300 amplitude elicited by positive alcohol expectancy vi stimuli and the degree of endorsement of negative/sedating expectancies on the selfreport measure. In addition, individual differences relating to alcohol expectancies were examined as well. These results were able to identify specific stimuli that violated expectancies for each individual, as well as those that tended to violate expectancies in systematic ways across subjects. These findings provide a way forward for more precise assessment and prediction based on the well developed cognitive model of Alcohol Expectancies. In sum, variations in the amplitude of the P300 were consistent with the model of Alcohol Expectancies. Words imputing negative/sedating effects of alcohol elicited a large P300 in individuals with higher positive alcohol expectancies. By indexing the brain's electrophysiological response sensitive to expectancy violations, these findings demonstrate concordance between verbal measures of alcohol expectancies, which by their very nature are introspective, and a psychophysiological index of expectancy thought to operate automatically and to be independent of overt responding.
590
Advisor: Mark Goldman, Ph.D.
653
Alcohol expectancies
Event-related potentials
P300
Semantic processing
PCA
690
Dissertations, Academic
z USF
x Psychology
Masters.
773
t USF Electronic Theses and Dissertations.
4 856
u http://digital.lib.usf.edu/?e14.3468



PAGE 1

Priming Expectancies: Effects on Neurophysiolo gical Indices of Exp ectancy Violations and Drinking Behavior by Ty Brumback A thesis submitted in partial fulfillment of the requirements for the degree of Master of Arts Department of Psychology College of Arts and Sciences University of South Florida Co-Major Professor: Mark S. Goldman, Ph.D. Co-Major Professor: Emanuel Donchin, Ph.D. Thomas Brandon, Ph.D. Date of Approval: 16 Feb 2010 Keywords: alcohol expectancies, event-rela ted potentials, P300, semantic processing, PCA Copyright 2010, Ty Brumback

PAGE 2

Acknowledgements I would like to thank my advisors, Dr. Mark S. Goldman and Dr. Emanuel Donchin for their support and guidance in my academic and re search pursuits. I am grateful to my committee member, Dr. Thomas Brandon, w ho provided great insight into the methodological challenges of examining s ubstance related cognition and behavior. Special thanks to Dr. Inna Fishman w ho laid the foundation for this project.

PAGE 3

i Table of Contents List of Tables iii List of Figures iv Abstract v Introduction 1 Priming and Alcohol Expectancies 3 Event Related Potentials 6 ERPs, Alcohol, and Affect 7 Study Rationale 10 Aims and Hypotheses 12 Method 13 Participants 13 Measures 14 Demographic Form 14 Oddball Task 14 ERP Expectancy Sentence Task 14 Alcohol Expectancy Questionnaire 16 Alcohol Expectancy Multi-Axial Assessment 16 Pattern of Alcohol Use 17 Thirty-day Timeline Follow-Back 17 Family History 18 Taste Rating Task 18 Procedure 18 Processing and Analysis 22 EEG Data Preprocessing 22 Principal Components Analysis 22 Statistical Analyses of Hypotheses 24 Results 26 Demographics 26 Expectancy and Drinking Variables 26 Taste Test Beverage Consumption 30 Oddball Task 35 Averaged Waveforms 35 Spatiotemporal PCA 37 Expectancy Sentence Task 41 Averaged Waveforms 41

PAGE 4

ii Expectancy and Priming Effects 41 Spatiotemporal PCA 46 Analysis of SF–TF Scores 49 SF3–TF3 50 SF3–TF1 50 SF3–TF5 52 ERP and Drinking Variables 55 Discussion 57 Prime – Expectancy Interactions Evident in ERPs 57 The “P300” 59 The P300 and Decisions to Use Alcohol 61 Conclusion 63 References 64

PAGE 5

iii List of Tables Table 1 Expectancy demogra phics by prime group and sex 28 Table 2 Drinking demographi cs by prime group and sex 29 Table 3 Cross-correlations between expectancy measures and drinking measures 31 Table 4 Correlations between dri nking variables and beverage consumption 33 Table 5 Correlations between expect ancy measures and beverage consumption 34 Table 6 Summary of regression analys es for models predicting SF–TF factor scores 54 Table 7 Correlations between dri nking and SF3–TF factor scores 56

PAGE 6

iv List of Figures Figure 1 Oddball averaged waveforms at Fz, Cz, & Pz electrode sites 36 Figure 2 Spatial factor loadings (v irtual electrodes) from PCA on Oddball data 37 Figure 3 Virtual ERPs for virtua l electrodes SF1, SF2, & SF3 39 Figure 4 Temporal factor loadin gs from PCA on Oddball data 40 Figure 5 Averaged waveforms at Fz Cz, & Pz electrode sites for sentence task 42 Figure 6 Prime groups compared between AEMax Woozy median split groups 44 Figure 7 Prime groups compared between AEMax Attractive median split groups 45 Figure 8 Spatial factor loadings from PCA 47 Figure 9 Virtual ERPs for SF2 ( central, right-lateral), & SF3 (centro-parietal) 48 Figure 10 Temporal factor loadings from PCA 49 Figure 11 Scatterplot of SF3–TF1 sc ores for Negative Alcohol Sentences by prime and AEMax Attractive 54 Figure 12 Scatterplot of SF3–TF1 sc ores for Negative Alcohol Sentences by prime and AEQ Gl obal Positive 55 Figure 13 Scatterplot of SF3–TF5 sc ores for Negative Alcohol Sentences by prime and AEMax Positive Arousing 55

PAGE 7

v Priming Expectancies: Effects on Neurophysiolo gical Indices of Exp ectancy Violations and Drinking Behavior Ty Brumback ABSTRACT Investigations of the anticipated eff ects of alcohol indi cate that cognitive frameworks are highly correlated with dri nking and other variable s associated with alcohol use, explaining up to 50% of th e variance in drinking outcomes (Goldman, Darkes, & Del Boca, 1999; Goldman, 2002; Goldman et al., 2006; Goldman, Reich, & Darkes, 2006). Furthermore, alcohol expectan cies appear to mediate the relationship between a variety of risk factors, such as sensation seeking, and alcohol outcomes (Darkes, Greenbaum, & Goldman, 2004). The cu rrent study examined the relationship of these cognitive networks with a physiol ogical index of expectancy violation Participants were presented with statem ents reflecting a wide range of alcohol outcome effects, which either violated or c onfirmed the participant’ s own set of alcohol expectancies, while the ERPs evoked by thes e stimuli were recorded. As predicted, the P300 amplitude elicited by negative alcohol e xpectancy stimuli was positively correlated with the degree of endorsement of positiv e/arousing expectancies on the self-report measure. That is, the higher the individual’s positive/arousing expectan cies, the larger the P300 elicited by stimuli asserti ng the negative effects of alco hol. There was no significant correlation, however, between P300 amplitude elicited by positive alcohol expectancy

PAGE 8

v istimuli and the degree of endorsement of ne gative/sedating expectancies on the selfreport measure. In addition, individual differences relating to alcohol expectancies were examined as well. These results were able to identify sp ecific stimuli that viol ated expectancies for each individual, as well as those that tended to violate expectancies in systematic ways across subjects. These findings provide a wa y forward for more precise assessment and prediction based on the well developed cogni tive model of Alc ohol Expectancies. In sum, variations in the amplitude of th e P300 were consistent with the model of Alcohol Expectancies. Words imputing negativ e/sedating effects of alcohol elicited a large P300 in individuals w ith higher positive alcohol expectancies. By indexing the brain’s electrophysiological response sensitive to expectancy violations, these findings demonstrate concordance between verbal meas ures of alcohol expectancies, which by their very nature are intros pective, and a psychophysiolo gical index of expectancy thought to operate automatically and to be independent of overt responding.

PAGE 9

1 Introduction Alcohol expectancies are anticipatory me mory processes that affect perception, cognition, and behavior related to alcohol. Investigations of the anticipated effects of alcohol indicate that these cognitive framewor ks are highly correlated with drinking and other variables associated with alcohol use, explaining up to 50% of the variance in drinking outcomes (Goldman, Darkes, & Del Boca, 1999; Goldman, 2002; Goldman et al., 2006; Goldman, Reich, & Darkes, 2006). For example, heavier drinkers tend to endorse more positive and arousing expectan cies compared to light drinkers, and expectancies measured in children prior to the onset of dri nking behavior are prospectively associated with drinking behavior (Goldman, 2002; Dunn & Goldman, 1998). Furthermore, alcohol expectancies app ear to mediate the relationship between a variety of risk factors, such as sensat ion seeking, and alcohol outcomes (Darkes, Greenbaum, & Goldman, 2004). This has led some researchers to consider alcohol expectancies to be one of the primary syst ems that accounts for biopsychosocial risk for alcohol use and abuse (Goldman et al ., 2006; Sher, Grekin, & Williams, 2005). Alcohol expectancies have been measur ed via semantic associations probed through direct self-report and indirect cogni tive paradigms (Goldman, Reich, & Darkes, 2006; Kramer & Goldman, 2003; Reich & Goldman, 2005). Although there is some debate over the utility of direct (explicit) versus indirect (implicit) measures, both classes of measures have explained variance in al cohol outcomes (Wiers et al., 2002, Reich, Below & Goldman, 2010). In fact, both overt and implicit manipulations of expectancies

PAGE 10

2 have led to changes in actual drinking be havior in several experiments (Darkes & Goldman, 1993; Roehrich & Goldman, 1995; Stein, Goldman, & Del Boca, 2000). In some respects, ascertaining expectancies vi a implicit measures provides insight that cannot be derived from explicit measures because implicit measures may reflect more automatic processing. Although implicit and expl icit measures of alcohol expectancies index the same construct to some degree, it is clear that implicit measures add predictive value incrementally to explicit measures (Reich et al., 2010). By harnessing this variance via implicit measures, researchers may be able to better explai n the anticipatory and predictive nature of expectanci es in the realm of alcohol and beyond in a way that moves beyond correlational designs to studies that show the effect of expectancies on information processing and decisi on making at the event level. Recently, Fishman, Goldman and Donchin (2008) developed a novel approach for implicitly measuring alcohol expectancies us ing event-related potentials (ERPs). They utilized an established brainwave paradigm th at measures violations of expectation and created an application in which stimuli would either f it with an individual’s alcohol expectations or violate one’s al cohol expectations. They then showed that the individual’s brain waves served as an index of the vi olation. This study provided evidence that expectancies predict responses to stimuli fa r more quickly than could be measured by language-based expectancy paradigms (i.e., w ithin milliseconds of stimulus presentation), thus substantiating the theory that expectancies serve as anticipatory frameworks for evaluation of stimuli encountered in the envi ronment. Furthermore, it opened the door for additional ERP investigations of alcohol expectancies at the level of individual differences.

PAGE 11

3 Priming & Alcohol Expectancies Apart from implicit measurement of al cohol expectancies, many studies have explored the implicit activation of alcohol e xpectancies. When activated, representations in memory trigger a host of related semantic and affective associations that can influence subsequent behavior (Bargh & Williams, 2006) As in classic priming experiments in which responses to a target are faster when a related word or concept has been primed, researchers have described num erous social and behavioral responses that result from subtle primes (e.g., Vohs, Mead, & Goode, 2006). Activation of concepts that facilitate these responses often occurs outside of one ’s awareness, yet this activation exerts measurable influence on subsequent behavi ors. In the alcohol expectancy domain specifically, researchers have capitalized on the fact that encountering alcohol cues activates associations an individual has w ith those cues, which includes everything from semantic associations to behavioral outc omes, potentially increasing the likelihood of using alcohol as a behavioral option (e.g., Leigh & Stacy, 1998). While direct cuing clearly has the potential to activate associations several studies ha ve used indirect priming of alcohol expectancies to examin e the influence of memory associations on subsequent activities. Several studies examined the effects of activation on actual drinking behavior. For example, one study examined the effects of two types of primes on female social drinkers’ behavior (Roehric h & Goldman, 1995). The primes included a video clip of a sitcom and a modified Stroop task, with a neutral condition and a positive alcohol condition for each type of prime. After view ing various combinations of alcohol and neutral primes, participants ha d the opportunity to consume placebo beer in an ostensibly

PAGE 12

4 unrelated study. Participants who viewed th e alcohol video clip and completed the alcohol Stroop task consumed more placebo beer than those in the neutral or mixed prime conditions. A similar study used the same St roop task priming paradigm, but added a negative alcohol condition in wh ich negative or sedating alco hol expectancy words were presented (Carter et al., 1998). Results indica ted that those primed with positive alcohol expectancy words consumed more beer in the taste test than those in the neutral condition and those primed with negative expectancy wo rds consumed less than those in the neutral condition. Another study compared the relatio nship between positive and neutral alcohol expectancy verbal primes and positive and neutral music mood inductions with subsequent beer consumption (Stein, Gold man, & Del Boca, 2000). Those primed with positive expectancy words drank more than thos e in the neutral condition, and within the positive expectancy group heavier drinkers consumed more than lighter drinkers. Several other studies examined the effects of priming on non-consumptive behaviors. For example, one study presented pa rticipants with a lis t of various alcohol expectancy and food words to study (Reic h, Noll, & Goldman, 2005). Primes consisted of two nearly identical word lists in which th e first word was altered to read “milk” in the neutral condition and “b eer” in the alcohol condition. Those in the “beer” condition tended to recall a greater pr oportion of expectancy words. Another study utilized a modified Stroop task in which participants we re primed with either neutral or alcoholic beverage words prior to the Stroop targets, which included arousing, sedating, or negative expectancy words, or neutral words (Krame r & Goldman, 2003). Results indicated that lighter drinkers demonstrated slower color naming reaction times to sedating expectancy

PAGE 13

5 words following alcohol beverage primes, wher eas heavier drinkers demonstrated slower reaction times to arousing expectancy wo rds following alcohol beverage primes. Cumulatively, these studies indicate that priming alcohol expectancies results in differing responses in cognitive performance and behavior, and that th e drinking status of the individual affects the results. That is, ac tivation of alcohol expectancies via contextual and language-based priming appears to faci litate or inhibit cognitive processing and decision-making differentially, indicating th at these memory networks may act as anticipatory mechanisms in subsequent stimulus evaluation and decision making, including the decision to drink alcohol. Behavi oral responses are automatically facilitated by indirectly priming alcohol expectancy wo rds and alcohol-related words, with the greatest facilitation coming from primes acco mpanied by a contextual cue (e.g., Roehrich & Goldman, 1995; Carter et al., 1998; Gold man, Darkes, Del Boca, 1999). These studies provide a framework within which to examin e alcohol expectancies; however, each of the studies above relies on behavior al output far downstream from the purported activation of alcohol-related concepts in memory. Newer approaches, such as the ERP paradigm developed by Fishman et al., are appealing because they allow researchers to explore the processes far upstream of those previously re ported. It is important to understand what ERPs are in order to understand how such a tool can contribute to the investigation of alcohol expectancies, and to understand how ERPs can build upon the strong empirical foundation of cognitive and semantic research established in the alcohol expectancy domain.

PAGE 14

6 Event Related Potentials Event-related potentials (ERPs) are compone nts of electrical ac tivity of the brain that are elicited by specific events. ERPs ar e time-locked to discrete sensory, motor, or cognitive events, and as such are manifestati ons of neural activiti es that are invoked in the course of information processing (for review see Fabiani, Gratton, & Coles, 2000). The ERP signal, which is only a few microvolts, is extracted from the electroencephgraphic (EEG) activity, wh ich can reach 50 microvolts, by signal averaging. The ERP waveforms reflect the eff ects of particular information processing elicited by the event. ERP methodology provide s a non-invasive tool with very fine temporal resolution (in milliseconds). ERPs have less spatial resolution to identify neural origins of electrical activity, though recent statistical techniqu es have allowed for source localization with greate r acuity than earlier me thods (Slotnick, 2004). The ERP elicited by an event consists of a sequence of components, labeled by polarity and latency in milliseconds (e .g., N100, P300). The activity that the ERP components manifest is assumed to have a func tional significance as sp ecified in terms of the information processing role of the underl ying neural action (D onchin & Coles, 1988). Early ERP components, with a latency of le ss than 100 ms, reflect sensory processes, while later components reflect higher cognitive processes like semantic processing and error monitoring (for a review see Ke y, Dove, & Maguire, 2005). To access the functional significance of ERPs, tasks must be designed to elicit specific information processing functions. For example, the P300 component is typically elicited using an “Oddball” paradigm in which participants are required to attend to a sequence of events in which are interspersed infrequent events The P300 is elicited by these infrequent

PAGE 15

7 events, while a component like the N400 is elicited using semantically incongruent terminal words of sentences (e.g., “I like my coffee with sugar and sand ”). The P300 component has been one of the most widely ex amined ERPs, due in part to the functional significance that is attributed to it. Several variables affect the amplitude and latency of the P300 component (see Fabiani et al., 1987; Picton, 1992). P300 amplitude increases and decreases as a function of stimulus probability and task relevan ce or value (Duncan-Johnson & Donchin, 1977). Furthermore, P300 amplitude is dependent on subjective probability and relevance of an event, while the latency of the P300 is largely dependent on task complexity (Donchin & Coles, 1988; Dien, Spencer & Donchin, 2003). T hus, subjective probability and relevance are important factors for understanding the implications of the P300 in cognition. The context-updating hypothesis pos its that unexpected events interrupt ongoing cognitive processes, causing the individual to revise the current model of the environment in working memory (Donchin, 1981; Donchin & Co les, 1988). Stimuli that are unexpected or that are most relevant to the individual require more si gnificant updating and result in larger P300 responses. The functional significan ce of the ERP components, particularly the P300, makes ERP a very powerful tool for accessing cognitive processes including those pertaining to the evaluation of alcohol-related information and alcohol consumption. ERPs, Alcohol, and Affect ERPs have been utilized frequently in examinations of the cognitive effects of acute and chronic alcohol consumption. The goal of most such studies has been to identify specific cognitive deficiencies associ ated with various levels of alcohol use by

PAGE 16

8 examining the amplitude and latency of specific ERP waves. Typically, tasks are contextually neutral and m easure simple information processing, such as pressing a button when rare stimuli are presented. While these tasks are valid for identifying information processing, they are non-specifi c because they do not have a special relationship to the phenomenon of interest : problem drinking. As will be delineated below, ERPs have the potential to access motivationally significant and emotionally relevant cognitions that pr ovide more functionally signi ficant aspects of perception, evaluation, and decision-making related to s ubjectively salient stimuli, including drugrelated stimuli. Just as priming paradigms have been used in behavioral research as mentioned above, such priming paradigms have proven effective in assessi ng context specific expectations and attitudes via ERPs. For exam ple, a mismatch between a primed affective category (e.g., good or bad; happy or sad) and a stimulus word, resulted in evaluative inconsistency and elicited a “late positive ERP component” (Cacioppo, Crites, & Gardner, 1996). Upon further evaluation this late positive “component” includes the P300 component when properly parsed (Ito & Cacioppo, 2007; Spencer, Dien, & Donchin, 2001; Dien, Spencer, & Donchin, 2004). Simila r results were found with affectively primed political attitudes (Mor ris et al., 2003). Furthermore, evidence indicates that these violations are automatic and uncontrollable, a nd may even conflict w ith reported attitudes in stereotype challenges (e.g., violations of gender stereotypes; Osterhout, Bersick, & McLaughlin, 1997). Individuals also exhibi t P300 to subjectively arousing picture stimuli, and larger responses appear to reflec t the level of affective arousal (Cuthbert et al., 2000). Thus, ERPs allow researchers acce ss to affect-laden and context-specific

PAGE 17

9 evaluative information processing, which may or may not be readily reported by the participant in direct self-report tasks. Alcoholand drug-related stimuli can access associated automatic cognitions in a similar manner to non-alcohol related studies cited above, though little research has been conducted on ERP responses to alcohol stimu li. Hansenne et al. (2003) examined ten alcoholics compared to controls and found a decreased P300 latency to alcohol-related words in the alcoholics, but no differences in amplitude. These results may be confounded by the preexisting attenuation of P300 response reported in alcoholics and their offspring (Begleiter et al., 1984; Begleite r et al., 1987). In studi es on drug stimuli, research has shown that drug-relevant stimu li increase ERP amplitude in those addicted to the drug (Franken et al., 2003). Therefore, context specific s timuli that were affectively salient to particular indivi duals elicited quicker and potentially larger ERPs. Another set of studies examined P300 res ponses to alcohol cues in drinkers with varying levels of sensit ivity to alcohol. They found that i ndividuals lower in sensitivity to alcohol elicited larger P300s to alcohol cu es and the P300 amplitude was correlated with self-reported drinking in th e following months (Bartholow, Henry, & Lust, 2007). This same group found that ERPs elicited by alc ohol cues correlated with self-reported positive evaluation of alcohol, whereas subjec ts’ self-reported evaluation of condom use appeared to conflict with ERPs to condom cues. While these studies utilize the idea that alcohol cues are automatically evaluated, it is necessary to understand the role of alcoholspecific cognitions in stimulus evaluation and decision making to connect the ERP literature seamlessly with alcohol literature.

PAGE 18

10 Fishman et al. (2008) was the first study to examine individual differences in P300 elicited by alcohol expect ancy statements. Participan ts were presented with statements about alcohol (i.e., “alcohol ma kes me…”) wherein the final word in each statement either agreed or c onflicted with the individual’s expectancies as indexed by a standard paper and pencil measure. Averaged waveforms indicated that individuals who primarily associated positive and arousing alcohol expectancies, which tended to be heavier drinkers, exhibited larger P300 res ponses to negative and sedating expectancy statements. Conversely, individuals who pr imarily associated negative and sedating alcohol expectancies, which were more likely to be lighter drinkers, tended to exhibit larger P300 responses to positive and sedating expectancy statements. That is, sentences that violated one’s primary expectancies e licited a larger P300 response than congruent sentences. Fishman et al. laid the groundwor k for the current study to attempt to expand the ERP paradigm they developed and to test whether this index of alcohol expectancy violation may also reflect i ndividual differences in salien ce or activation within an experimental task. Study Rationale Expectancies are assumed to be stored in memory as templates of systematic relationships between contextual cues and outcomes. They are designed to allow an individual to deal w ith the environment, and as such must be constantly updated to remain efficient (Goldman, 1999, 2002). As re viewed above, functional aspects of ERP research fit well into research on alc ohol related cognitions. The P300 component responds to subjective probability and can be used to assess the expectations of individuals. The P300 component appears to index the level of expectancy violation (i.e.,

PAGE 19

11 stimuli with the lowest probability appear to elicit the largest P300 in studies in which the probability of stimuli was manipulated). Rese arch has also shown that the P300 can be elicited with context specific and affect-lad en stimuli that assess specific attitudes and indicate violations of expected outcomes, a nd an initial study of th e relationship between alcohol expectancies and ERP responses indicated that the P300 is a valid correlate of violations of alc ohol expectancies. Research on language-bas ed assessments of alcohol expectancies has demonstrated reliable intensification through indirect priming and direct activation of alcohol memory networks that has affect ed memory performance, discrimination performance, and actual drinking behavior (Reich, Noll, & Goldman, 2005; Roehrich & Goldman, 1995). The P300 response observed in Fishman et al. (2008) may index alcohol expectancy violations, but it is unknown whether priming alcohol concepts prior to testing may augment such responses. Expectancy theory would predic t that activation of anticipatory cognitive frameworks affects perception and categorization of stimuli, and if the P300 is sensitive to this ac tivation of expectancy it should reflect this in increased or decreased amplitude. In order to further assimilate these ER P findings into the alcohol expectancy framework, it is necessary to examine wh ether the neurophysiological indexing of expectancy violations by the P300 com ponent also responds to experimental manipulations like other cognitive and behavi oral measures of expectancy. Previous studies have purportedly increased or decrea sed expectancies via priming, though it is unclear whether underlying expect ancies were changed or wh ether the effects elicited

PAGE 20

12 were a function of expectancy activation in a particular contex t-dependent state (e.g., Stein, Goldman, & Del Boca, 2000). The current study was designed to test whet her indirectly primi ng alcohol affected the amplitude or latency of the P300 component in an ERP paradigm, and then examined how the ERP responses related to other cogni tive and behavioral measures of alcohol expectancies. This design allowed for the eval uation of potential mechanisms of action of alcohol expectancies by expl oring whether the neurophysiol ogical activity following expectancy activation correlates with drinki ng behavior. By explor ing the effects of priming on neurophysiological measures of expectancy violation and on subsequent behavioral measures of exp ectancy activation, the study was designed to examine the role of perception and stimulus evaluation in expectancies’ effects on behavior. Aims & Hypotheses The study utilized an indirect priming manipulation in order to activate alcohol expectancy memory networks prior to the presentati on of stimuli in an ERP paradigm. It was expected that the ne utral prime condition would replicate results obtained in an earlier investig ation of ERP indices of alco hol expectancy violations (Fishman et al., 2008). That is, individuals who endorse more positive and arousing alcohol expectancies were exp ected to show larger P300 resp onses to incongruent (i.e., negative and sedating) alcohol statements. In the alcohol prime condition ERP results were expected to be moderated by expectancies (and drinking status of th e individual inasmuch as drinking and expectancies are correlated). Previous studi es on expectancy activation showed effects for most college students, as activating alc ohol related concepts tends to increase the salience of other alcohol associations. In this case, violations of expectancy were being

PAGE 21

13 examined, so individuals who endorse more positive and arousing alcohol expectancies were expected to react differently than individuals who endorse more negative and sedating expectancies. Following the alcohol prime, it was pred icted that indi viduals with positive and arousing expectancies would exhi bit a larger P300 response to negative and sedating alcohol sentences compared to t hose in the neutral prime group. Given the ambiguous results from the initial study in i ndividuals with more negative and sedating expectancies, coupled with the fact that individuals with a ny experience with alcohol tend to have some positive alcohol expectancies (Rather et al., 1992) no hypothesis was made concerning the ERPs exhibited by indivi duals with more negative and sedating expectancies. It was expected that alcohol expectanci es would correlate with reported drinking levels (i.e., higher positive a nd arousing expectancies correla te with heavier drinking), and that heavier drinkers and individuals w ith greater positive and arousing expectancies would consume more during an ad lib drinki ng session. It was exp ected that the group that viewed the alcohol vide o prime would drinking more in the ad lib drinking session after controlling for drinki ng levels. In addition, it was also hypothesized that ERP responses would be associated with subseque nt ad lib drinking (i.e., larger P300 to negative/sedating stimuli would correlate with greater drinking). Method Participants College students aged 21 and up were recruited through the university’s online research participant pool. Participants we re randomly assigned to neutral and alcohol prime conditions, with equal numbers of males and females in each condition. The

PAGE 22

14 language-based tasks coupled with EEG reco rding require several restrictions to participation. Participants were screened via an online demographic survey associated with the research part icipant pool, in which individuals were required to endorse having consumed alcohol in the last month, being na tive English speakers, and having normal or corrected-to-normal vision. They were also scre ened for history of neurological disorder (e.g., seizure disorder or multiple sclerosis) or head injury (i.e., loss of consciousness > 5 min), which could affect the EEG quality, as well as for use of medications that might affect EEG signal (e.g., anxiol ytics or neuroleptics). Measures Demographic form This form provided information regarding age, gender, ethnicity, education, and health status (specifically history of head injuries, neurological disorders, and current medication). Oddball task A standard oddball task with X and O stimuli was used. A total of 200 trials are included with 40 “rare” targets and 160 standard stimuli. Each stimulus was presented for 600 ms and the intra-stimulus-int erval (ISI) was set to 1000 ms. Participants responded to each trial by pressing one button following an X or another button following an O. ERP expectancy sentence task The ERP stimulus set used by Fishman, Goldman, and Donchin (2008) was utilized. Th is paradigm consists of 72 English statements describing various habits or act ivities pertinent to the college students, including studying, spending time with peer s, partying, drinking, smoking, exercising, etc. Each statement was missing the last wo rd, e.g., “On a Friday night, alcohol makes me....” and participants were instructed to pr ess a key to move to the next screen. When

PAGE 23

15 the key was pressed, a fixation point appeared for 500ms followed by the last word (e.g., “happy”) for 800ms. These targets words were chosen from the Alcohol Expectancy Multiaxial Assessment scale (AEMax), which contains 132 most common alcohol expectancy words derived by various item se lection procedures from a large pool of responses to the prompt “Alcohol makes one …” and subsequently normed in large college student samples, as described by Goldman and Darkes (2004). In total, 31 sentences related to alco hol: 14 with a negative/sed ating ending, and 17 with a positive/arousing ending (e.g., “Alcohol makes me… happy” vs. “Alcohol makes me…sad”), in a semi-random order. Fifteen sentences were structurally similar statements, but related to smoking (e.g., “S moking makes me…sick”), with 8 positive and 7 negative endings. These statements were borrowed from the Smoking Consequences Questionnaire (SCQ; Brandon & Baker, 1991). Another 17 sentences were composed with other, non-alcohol or nonsmoking content, such as exercising or studying (e.g., “After a workout at the gym, I always feel…exhausted”). They were intended as control/neutral condition for the ERP comparison. Finally, 9 classic N400eliciting sentences (e.g., “I drink my coffee with sugar and…socks”) were included in order to control for participan ts’ attention to the task. The 72 sentences made up 6 experimental conditions: AlcoholPositive/Arousing, Alcohol-Negative/Sedating, SmokingPositi ve, SmokingNegative, Incongruent, and Other – a baseline condition. One block of these 72 sentences was engineered so that no two statements from the same category were allowed to appear in a row, and no more than two statements of the same valence – e.g., positive alcohol expectancy – could follow each other, even if separated by f iller items. This block was presented first,

PAGE 24

16 followed by a block of all 72 sentences pr esented in random order determined by a computer that differed for each participan t. The P300 amplitude elicited by AlcoholPositive/Arousing and Alcohol-Negative/Sedating stimuli served as the main outcome measure for this task. Alcohol Expectancy Questionnaire (AEQ; Brown, Goldman, Inn & Anderson, 1980; Brown, Christiansen & Goldman, 1987; Goldman, Greenbaum & Darkes, 1997). The measure included 68 statements in a True /False format about the various effects of alcohol, including social, physic al and sedating domains. Expectancy items on the AEQ correlate with alcohol consum ption, alcohol abuse and beha vior while drinking, with a mean reliability of 0.84. Factor analysis rev ealed 6 separate subscales within this measure, including: global positive changes, sexual enhancement, physical and social pleasure, increased social asse rtiveness, relaxation and te nsion reduction, and arousal and aggression. The relative levels on each subs cale were analyzed to provide further information into the type of alcohol expect ancies endorsed by each participant. The AEQ was completed by participants through the onlin e participation program prior to enrolling in the study. Alcohol Expectancy Multi-Axial Assessment: Short Form (AEMax; Goldman & Darkes, 2004). The shortened version of th is measure included 24 expectancy items, with three from each of the eight first orde r factors derived from the longer 132-item scale (i.e., horny; social; egotistical; attract ive; sick; sleepy; woozy; and danger). These eight first order factors load onto three hi gher order factors: Positive-Arousing, Sedating, and Negative. Participants were asked how often they believe the item best completes the sentence “alcohol makes one…”, using a 7-poi nt Likert Scale rangi ng from 0 = “never”

PAGE 25

17 to 6 = “always”. The measure is proven bot h reliable and valid a nd is an effective measure of the positive-negative and ar ousing-sedating dimensions of alcohol expectancies. While many of the words overlap with those in the ERP task, this measure will give an explicit index to c ontrast with the ERP results. Pattern of alcohol use Participants were asked to re port their drinking habits for the past year. This includes the frequency and quantity of their typical alcohol use, the number of occasions on which they become dr unk from alcohol, as well as an item which queries how often they consumed beer. Rega rding the veracity of self-reports, the relevant literature indicate th at verbal reports can provide reliable and valid information when inquiries are made about sensitive personal information such as alcohol consumption, especially under circumstances in which there are no obvious incentives to underor over-report (see Babor, Brown, & Del Boca, 1990; Del Boca & Noll, 2000). Family Grid Family history has been identified as an important factor in the development of alcohol used di sorders as well as in respons es on behavioral and psychophysiological measures (e.g., Schuckit et al., 199 2; Porjesz et al., 2005). The family grid interview measures the density of first a nd second degree biologica l relatives having in the past or currently having significant drinking problems. Problems are assessed by endorsing items in one or more major life ar eas including: legal pr oblems (drunk driving violations), health problems (c irrhosis of the liver, alc ohol withdrawal), relationship problems (objections about drinking from fa mily members), work or school problems (absenteeism, poor performance due to alcohol use), and actual tr eatment (detox, rehab, AA meetings).

PAGE 26

18 30-Day Timeline Follow-Back (TLFB; Sobell & Sobell, 1992). This calendarbased interview measures participant alcohol use (quantity and freque ncy) retrospectively over the month prior to assessment. Participan ts were asked to identify the amount of alcohol consumed per drinking day in the pr evious month, with special attention to drinking patterns in the previ ous week. At the conclusion of the interview, participants were asked whether the calendar represents a typical drinking month. If the month was not considered typical, partic ipants were asked whether th e prior month shows a heavier or lighter drinking pattern. Taste-rating task The drinking session of the study was purported to be a market research taste test. Protocol from previous studies was used (Roehrich & Goldman, 1995; Marlatt, Demming, & Reid, 1973). Pa rticipants were asked to pa rticipate in a taste-testing study which required tasting a nd rating beverages on various characteristics including taste, texture, and color. The beverages incl uded separate brands of non-alcoholic beer to preclude the use of time consuming and cos tly protective measures required by NIAAA for alcohol consumption research, including pr egnancy tests for females. Participants were presented with 3 glasses of beer, with 8 ounces (~236 ml) in each glass, and were given a rating form with various characteris tics to be evaluated as well as an overall rating for each of the three samples. Each participant spent 7-10 minutes in the tasterating task. The main dependent variable was the total amount of the beverage consumed. Procedure Individuals were recru ited through the online part icipant pool based on the inclusion criteria detailed above. Eligible participants were invited to attend a 1.5 hour lab session in exchange for clas s credit. Participants were aske d to refrain from alcohol or

PAGE 27

19 non-prescription drug use for 24 hours prior to their appointments, to eat 4-6 hours prior to their appointment, and to refrain from st renuous exercise for at least 3 hours prior to their appointment. When participants arrived at the lab they were asked to complete an informed consent form, which provided information on confidentiality, bene fits and risks of participation, and storage of the data. Th e title of the study wa s listed as “ERPs & Memory Function” and the consent form did not provide information about the specific stimuli that would be viewed during the study to prevent participants ’ knowledge that the study would contain information about alcohol. Participants were also given the basic agenda of the experiment and were told th at there would be a 10 minute break at one point, during which they could have the option of participating in a market research study in a separate lab in the ps ychology department. After completing the consent form, the participant filled out a basic demographic form to confirm eligibility and gather additional information. Upon completion of th e demographic form, the EEG sensor net was placed on the participant’s head and the participant was led into an adjacent room where the EEG tasks were completed. First, a standard “Oddball” task was admi nistered in which participants responded by pressing a button when a ta rget letter was presented (“ X” or “O”), and pressing a different button when a non-targ et letter was presented. There were a total of 200 trials with targets presented 80% of the time and non-targets 20% of the time. This task served as a baseline for the participant’s individual response amplitude and latency, as well as a potential index of the genera l cognitive differences previous ly observed between at-risk and low-risk drinkers in P300 paradigms. Each ERP task bega n with a practice block to

PAGE 28

20 ensure the participant underst ood the instructions and was able to follow the directions. The experimenter remained in the room dur ing this period and guided the participant through this portion of each task and left the room during the recording phase. After completing the Oddball task, partic ipants viewed two short video clips presented under the guise of a memory study. Participants continued to wear the EEG sensor nets, though EEG was not being recorded during the video clip presentation. Each of the clips was approximately 90 seconds long. The first video clip depicted a conversation among friends in a coffee shop (from the sitcom Friends ) and was presented in both conditions. The second video clip di ffered between conditions. In the neutral condition the clip was depicts a conversation among friends in a diner (from the sitcom Seinfeld ). In the alcohol condition, the second cl ip also depicted a conversation among friends, though the setting of this dialogue is a bar (from the sitcom It’s Always Sunny in Philadelphia ). The two videos that differed between conditions were matched on several variables including number of speakers, level of humor, and level of arousal, in an attempt to make the setting of the di alogue the main distinguishing feature. After a short rest during which the experi menter ensured that the participant was comfortable and that the electr odes were still reading properly, participants completed the expectancy violation task. For the expectan cy violation task, ERP recording were timelocked to the onset of the final word in each sentence and the recording epoch of each trial was 1000ms. At the offset of the target word participants were asked to perform a judgment task (Do you agree/ disagree with th e statement?). They were instructed to make their response as soon as the prompt a ppeared on the screen and reaction times of

PAGE 29

21 this response were recorded. After the part icipant’s response, th ere was a 2000-ms intertrial interval before the next trial began. Following the ERP tasks, the sensor net was removed from the participant’s head and the participant was given the option of par ticipating in a short ma rket research study during the 10 minute break of the experime nt. When participants agreed, a second experimenter took them to a lab down the ha ll from the EEG lab where they were told they would be participating in a taste-rating task. Participan ts completed a consent form and demographic form to uphold the distinct ion between the EEG experiment and the taste-test task. Participants were presented with three gla sses of non-alcoholic beer and a glass of water to rinse their mouths between beers. They were given brief instructions then left alone with a rating form for each beverage. The experimenter running the tastetest checked in on the participan t after 5 minutes, and if the participant indicated that they were done the experimenter delayed and return ed to retrieve the participant 2 minutes later. If the particip ant was not done at 5 minutes, the ex perimenter returned to the room at the end of 10 minutes. Particip ants then filled out the patte rn of alcohol use survey in the same room, with instructions that the res earchers running the tastetest wanted to take into account how often people typically drink. After completion of the taste te st portion of the study, the part icipant was led back to the EEG lab preparation room to complete the remaining questionnaires (AEQ, and AEMax) followed by the TLFB and family history interv iews with the initial experimenter. The participant was debriefed, during which knowle dge of the experimental manipulation was queried and the nature of the study and the contents of the beverages were then disclosed.

PAGE 30

22 Data Processing & Analysis EEG data preprocessing EEG data were digitally filtered with a 40-Hz lowpass filter and segmented into epochs starting 100 ms prior to stimulus onset to 1000 ms following stimulus onset for the Expectancy Sentence paradigm, and 200 ms prior to stimulus onset to 1000 ms following stimulus onset in the Oddball paradigm. These raw EEG epochs were then run through an automate d artifact detection program, corrected for eye movements (Gratton, Coles, & Donchin, 1983), and baseline-corrected using the average of the respective pre-stimulus epoch. Th e artifact-free trials were then averaged separately for each experimental condition. In the Expectancy Sentence paradigm this yielded 6 separate average waveforms (i.e., Alcohol-Positive/Arousing, AlcoholNegative/Sedating, SmokingPositive, Smoki ngNegative, Incongruent, and Other) for each participant. In the Oddba ll task this sequence of pr ocessing steps generated two separate average waveforms for each partic ipant; one for rare “targets” and one for frequent “standard” stimuli. Finally, the aver aged data were re-referenced to a meanmastoid reference. This procedure generate d a 129th channel of mathematically linked reference recorded separately from the ear l obes. Participants with fewer than 75% good trials per category were exclude d (n = 7). The averaged data files were then exported to MATLABTM (version 2008a, MathWorks, Natic k, MA) for further processing. Principal Components Analysis In order to extract components a principal components analysis (PCA) was conducted on the observed waveforms. The extracted components were not based on peaks or troughs in the raw waveform but on the basis of experimental variation. A software packag e called PCA Toolbox (version 1.22; Dien, 2008), was utilized to run the PCAs in MATLAB.

PAGE 31

23 The PCA procedure required several ma thematical steps that are completed automatically by the software package. To begin with, correlati ons are calculated between each electrode pair over all the time points. The PCA procedure then forms combinations of the original measures that capture the most relevant variance. Each principal component is a weighted linear co mbination of all the original dependent variables. PCA is intended to describe th e complex relations between the many variables in terms of a smaller numbe r of hypothetical, unobserved, latent variables that do not overlap significantly. These components reflect “some essential physiological, psychological or hypothetical construct whose properties are under study” (Donchin et al., 1977, p. 10). The principal components are extr acted from the data in a hierarchical fashion. The first component accounts for the largest proportion of the variance in the data, and the successive components must accoun t for the largest portion of the residual variance. Using an orthogonal rotation (e .g., Varimax) forces each component to be uncorrelated, but when using an oblique ro tation (e.g., Promax) the solution begins with the orthogonal components (i.e., simple struct ure solution matrix) and then rotates the solution by seeking a least squares fit so so me of the components end up being correlated. For typical ERP data, this percentage of vari ance accounted for drops off rapidly after the first five or six components, which usua lly account for 90–95% of the variance in the data. The components extracte d are thought to represent th e variance controlled by the experimental manipulation (in the case of th e P300, the degree of expectancy violation). To derive the P300 component several steps were required. In ERP data, the variables are the microvolt readings at each electrode and at each consecutive time point. A spatial PCA was conducted for the averaged waveforms at each

PAGE 32

24 electrode site for all experimental conditions for each participant, to reduce the number of variables in this dimension. Spatial PCA identifies clusters of electrodes that are so highly correlated that so me of the electrodes can be cons idered redundant (Spencer, Dien, & Donchin, 2001). The spatial PCA produce a series of “spatial factors” from the original 129 electrodes that represent hi ghly correlated electrodes. After reducing the dataset to a set of spatial factors, a temporal PCA was conducted to reduce the temporal dimensions. In this step, the spatial factor scores associated with the time points of the original dataset were used as the variables for the PCA, and the observations were the spatia l factors. The resulting spatiotemporal factor scores (i.e., scores for a given spatial factor at a given tem poral factor) then served as dependent variables for subsequent analyses Specifically, a combination of the spatial factor accounting for the most variance in the centro-parietal channels (corresponding to the well-established scalp distri bution of P300) and the tempor al factor accounting for the most variance in the window corresponding to the P300 latency (300-600 ms) were sought to represent the P300 ERP co mponent as a dependent variable. Statistical analyses of hypotheses Demographic and drinking data were first examined to determine whether the random a ssignment procedure resulted in groups that did not differ on any of the variables of intere st (e.g., drinking and expectancy variables). Drinking variables (derived fr om quantity/frequenc y self-report items and the TLFB) and expectancy variables (AEQ and AEMax) were then correlated to confirm the presumed association between drinking leve ls and expectancy levels fo r further analysis. Analyses were then conducted on the amount of beer c onsumed in the taste-test portion of the experiment. That is, correlations between expectancy variables and the amount of

PAGE 33

25 beverage consumed (in milliliters) were examin ed in the entire sample and in each prime group separately. Beverage consumption was also compared between priming groups to explore the effect of the prime condition on subsequent drinking. ERP data were then examined first to determine whether spatiotemporal factor scores, believed to represent the P300, reflect alcohol expectancy vi olation as presented by Fishman, Goldman, and Donchi n (2008) in the ini tial study. Thus, in order to examine the hypothesis that sentences describing al cohol effects that are deviant from a participant’s subjective set of alcohol expect ancies will elicit a larger P300, expectancy scores were correlated with the spatiotempor al factor scores for responses to AlcoholPositive/Arousing and Alcohol-Negative/Sedating items. This was done first for the neutral prime group. It was predicted that individuals with higher positive/arousing expectancies (i.e., heavier dr inkers) would respond with a larger P300 to statements describing negative/sedating eff ects of alcohol consumption (i.e., a positive correlation with Alcohol-Negative/Sedating factor sc ores) indicating that the sentences were unexpected or less congruent with their exp ectancies. The converse hypothesis was also explored, though it was unclear whether i ndividuals with higher negative/sedating expectancies (i.e., light er drinkers) should have been exp ected to produce a larger P300 to positive/arousing expectancies as previous results were not significant. The extracted P300 component was then compared in the alcohol priming condition in a similar fashion. These data were then entered into a seri es of regressions, using the correlation matrices as guides as to which expectancy measures and drinking variables to include. Multiple regressions were constructed using the spatiotemporal factor scores for Alcohol-

PAGE 34

26 Positive/Arousing and Alcohol-Negative/Sedating as the dependent variable and an expectancy variable, primi ng group, and the expectancyby-prime interaction. This allowed the examination of the hypothesized main effects or expectancy levels, as well as the potential interaction between prime and expectancy, which was exploratory in nature. Results Demographics Sixty-two participants were recruited, and 55 participants comprised the final sample because data had to be excluded from seven participants due to excessive artifacts in the EEG recordings (attributed to factors such as eye or other muscle movement and difficulty in net application due to thick hair ). Three additional participants refused to participate in the market research portion of the study. Their data are included in the ERP analyses, but are excluded for analyses pertai ning to the beverage consumption portion of the study. Twenty-four individuals participated in the Alcohol prime condition (12 males/ 12 females), and 31 participated in the Neut ral prime condition (15 males / 16 females). Seventy-five percent of the sample was ag ed 21-25, 18% was aged 26-29, and 7% of the sample was aged 30-41. Seventy-three percen t of the sample reported being white/nonHispanic, 9% Hispanic/Latino, 7% Asian, 6% black, and 5% reported being “other”. Expectancies and Drinking Variables Table 1 lists the means and standard devi ations of the self-report expectancy measures by prime group and sex. As expecte d, males reported higher expectancies than females on most subscales. There were no significant differences overall between the prime groups, and comparisons by sex between the prime conditions revealed no significant differences either.

PAGE 35

27 Table 2 lists the results of drinking va riables derived from the Timeline Followback measure as well as from the single-item quantity and frequency measure collected during the market research por tion of the study. No significan t differences exist between the sexes within prime condition, between prime conditions within sex, nor overall between prime conditions for the TLFB derived variables. The mean “Drinks in last month” for the males in the Alcohol prime group varies widely and is inflated due to an outlier who was greater than 3SD above the mea n, but that participant’s data was left in the dataset since the means between males compared across prime condition were not statistically signifi cantly different [ t (25) = -1.59, p =.12]. For the quantity/frequency measures, males reported drinking higher quantities per occasion than females [ t (50) = 3.44, p <.001]. Males also tended to endorse drinking beer more frequently than females [ t (50) = 2.12, p <.05; non-significant after Bonferroni adjustment for multiple comparisons]. The Alcohol Prime group tended to report drinking beer more than the Neutral Prime group [ t (50) = -2.45, p <.05], and specifically males in the Alcohol prime condition tende d to report drinking beer more frequently than males in the Neutral prime condition [ t (25) = -2.25, p <.05], though these effects were again only a trend when Bonferroni’s adjustment was used for multiple comparisons. These results are important to take into account when in terpreting the results from the beverage consumption portion of the study detailed below.

PAGE 36

28 Table 1 Expectancy Demographics by Prime Group and Sex Alcohol Prime Group (n = 24) Neutral Prime Group (n = 31) Males (n= 12) Females (n= 12) Males (n= 15) Females (n= 16) Alcohol Expectancy Questionnaire AEQ – GP 9.17 (6.0) 6.58 (5.1) 7.87 (3.9) 7.63 (5.6) AEQ – SA 7.17 (2.1) 7.00 (1.5) 7.20 (2.5) 6.69 (3.5) AEQ – Agg 4.75 (2.6) 5.17 (2.5) 4.20 (2.6) 4.31 (2.5) AEQ – Sex 1.92 (1.9) 2.50 (2.2) 2.47 (2.5) 2.13 (2.2) AEQ – SPP 7.58 (1.2) 7.50 (1.8) 7.53 (1.5) 7.19 (1.7) AEQ TR 5.33 (2.2) 5.50 (3.1) 5.33 (2.1) 6.25 (3.0) AEQ – Tot 35.92 (13.6) 34.25 (12.5) 34.60 (11.2) 34.19 (15.4) Alcohol Expectancy Multiaxial Assessment AEM – Horny 9.58 (4.0) 8.50 (3.7) 10.13 (2.6) 7.69 (3.7) AEM – Ego 10.00 (3.6) 7.25 (5.2) 9.07 (3.6) 7.25 (5.3) AEM – Sick 8.50 (3.2) 7.92 (3.3) 7.80 (2.5) 7.19 (4.4) AEM – Woozy 8.67 (3.3) 8.58 (3.3) 9.73 (3.1) 8.13 (3.8) AEM – Social 14.08 (2.2) 12.75 (2.7) 13.60 (2.0) 13.06 (2.9) AEM – Attract 8.33 (3.7) 7.42 (3.5) 7.73 (3.0) 7.38 (4.6) AEM – Sleep 10.25 (4.0) 8.75 (4.6) 9.67 (3.3) 9.06 (2.7) AEM – Danger 7.50 (3.6) 6.17 (4.8) 5.73 (4.5) 6.13 (6.0) AEM – Sed 27.42 (9.3) 25.25 (9.4) 27.20 (6.2) 24.38 (8.9) AEM – Neg 17.50 (6.0) 13.42 (9.3) 14.80 (7.4) 13.38 (10.8) AEM – PA 32.00 (7.7) 28.67 (8.8) 31.47 (5.5) 28.13 (9.4) Note. Data are Mean (SD). AEQ – GP = Global Positive; AEQ – SA = Social Assertion; AEQ – Agg = Aggression; AEQ – Sex = Sexual Enhan cement; AEQ – SPP = Social & Physical Pleasure; AEQ – TR = Tensi on Reduction; AEQ – Tot = Total of all 68 items. AEM – Sed = Sedating higher order factor; AEM – Neg = Negative higher order factor; AEM – PA = Positive & Arousing higher order factor.

PAGE 37

29 Table 2 Drinking Demographics by Prime Group and Sex Alcohol Prime Group (n = 24) Neutral Prime Group (n = 31) Males (n= 12) Females (n= 12) Males (n= 15) Females (n= 16) Timeline Follow-back Drinking Variables Drinks in last month 30.67 (24.8) 18.25 (15.4) 18.67 (13.8) 20.63 (21.2) Days Drinking in last month 7.00 (4.1) 6.17 (5.6) 5.13 (3.6) 5.88 (4.7) Drinks / Drinking Day 4.10 (1.6) 3.28 (1.7) 3.43 (1.7) 2.74 (1.4) Drinks per Week 6.58 (5.2) 4.02 (3.2) 4.10 (3.0) 4.51 (4.7) Quantity/Frequency Variables (n= 15) (n= 13) Frequency in last year 5.17 (1.4) 5.00 (1.7) 4.73 (1.4) 4.85 (1.6) Quantity per occasion 4.33 (1.2) 3.00 (1.1) 3.93 (1.7) 2.77 (1.1) Drunk occasions in last year 3.92 (1.9) 3.08 (1.7) 2.93 (2.1) 2.77 (2.1) Frequency of Beer drinking* 5.00 (1.3) 3.50 (2.5) 3.33 (2.3) 2.08 (2.1) Statistical trend of a difference be tween Alcohol Prime and Neutral Prime Table 3 details the correlations betwee n expectancy measures and drinking measures for the entire sample. Overall, the strongest relationships between drinking measures and expectancy measures were negative relationships between the AEMax Woozy subscale and TLFB deri ved drinking variables, as well as between the AEMax Sedating higher order factor and TLFB derived drinking vari ables. These relationships indicate that this sample is likely comprised of lighter drinkers than some other college sample studies. These relationships may also explain the lack of correlations between AEQ variables and drinking, since the AEQ i ndexes only positive expectancies which are

PAGE 38

30 apparently less predictive of drinking in this sample. A few positive relationships were present between the AEMax Attractive and th ree of the TLFB derived drinking variables as well as the single item Frequency measur e. In addition the AEMax Positive /Arousing higher order factor exhibited a positiv e relationship with the single-item quantity/frequency variables a nd the frequency (Days drinking) from the TLFB. Most of the correlations were in the expected dire ction (positive and arousing expectancies would be positively related to drinki ng and negative expectancies w ould be negatively related to drinking), but were not si gnificant. Relationships betw een the AEQ and AEMax were correlated as expected (not s hown). Apart from the strength of negative expectancies in this sample, it is unclear w hy many of the AEQ variables were not significantly or positively correlated with drinking variab les. A potential methodological factor contributing to this effect is that the AEQ was completed on the computer up to one month prior to completing the in-lab portion of the study so it was less proximate and had fewer controls on extraneous contextual f actors that may have influenced responses. Taste-test Beverage Consumption Fifty-two participants comp leted the taste-test por tion of the study (28 from Neutral Prime group; 24 from Alcohol Prime group). As expected, males (202 ml) tended to drink more than females [123 ml; t (50) = 1.78, p = .08]. In addition, several drinking variables from the TLFB and from the drinking profile were positively correlated with the amount of beverage consumed (see Tabl e 4). There were no differences in mean consumption between the Alcohol (154 ml ) and Neutral prime (173 ml) groups [ t (50) = .41, p = .68]. Therefore, the alcohol prime di d not appear to increase consumption directly; however, the relationships be tween drinking variables and beverage

PAGE 39

31 Table 3 Cross-Correlations between Expectancy Measures and Drinking Measures Expectancy measures Drinks in last month Days Drinking in last month Drinks / Drinking Day Drinks per Week Frequency Quantity per occasion Drunk Frequency of Beer drinking AEQ – GP .17 .31* -.05 .18 .21 .17 .15 .39** AEQ – SA .10 .16 .04 .10 -.07 .22 .07 .27 AEQ – Agg .10 .21 -.03 .13 .17 .11 .15 .25 AEQ – Sex -.07 .08 -.25 -.05 -.16 .02 -.23 -.05 AEQ – SPP .12 .21 -.06 .13 .05 -.05 -.15 .10 AEQ – TR -.03 .15 -.25 -.02 -.00 -.08 -.21 .12 AEQ – Tot .10 .26 -.12 .12 .08 .11 .00 .28* AEM – Horny .12 .22 -.10 .18 .24 .34* .31* .20 AEM – Ego .13 .17 -.05 .14 .26 .17 .27* .07 AEM – Sick -.33* -.29* -.22 -.32* -.15 -.01 .06 -.16 AEM – Woozy -.50** -.51** -.28* -.49** -.36** .03 .01 -.30* AEM – Soc .12 .27* -.18 .14 .05 .04 .05 .13 AEM – Attract .26* .38** -.01 .28* .37** .22 .11 .32* AEM – Sleep -.39** -.41** -.19 -.38** -.26 .01 -.16 -.12 AEM – Danger -.23 -.28* -.08 -.21 -.13 .03 .03 -.09 AEM – Sed -.51** -.50** -.28* -.49** -.32* .01 -.04 -.24 AEM – Neg -.06 -.06 -.07 -.04 .07 .10 .16 -.01 AEM – PA .21 .36** -.11 .25 .30* .27* .21 .28* p <.05 ** p <.01

PAGE 40

32 consumption were stronger in the Alcohol Prime group than in the Neutral Prime group. The relationships were all in the expected positive direction, a nd several variables exhibited significant relationships (Table 4). In addition to th e drinking variables, several expectancy variables includi ng AEQ Global Positive, Social and Physical Pleasure, Aggression/Arousal, and the Total score exhi bited positive relationships with the amount of beverage consumed, as did the Horny s cale from the AEMax (Table 5). A similar phenomenon occurred in the AEQ scores as in the drinking variables, in that the Alcohol Prime group exhibited stronger relationships w ith beverage consumption than the Neutral Prime group. The differences between the Alcohol Prime group and the Neutral Prime group were examined to determine whether the prime condition moderated the effect of preexisting expectancies and reported drinking le vels on the amount of beer participants consumed in the ad lib drinking session. Th e beverage consumption variable was first transformed using the natural log since it wa s not normally distribu ted and a series of ANOVAs were conducted examining the main effect of Prime as well as potential interactions with drinking and AEQ variables. These models suggested that most of the variance in beverage consump tion is explained directly by the expectancy and drinking variables, as none of the in teractions between prime and the identified variables was significant. The current sample was dreadfully underpowered for examining differences in such relationships, so while it appears that the alcohol pr ime influenced responses in a way that led to stronger c oncordance of self reported measures of drinking and

PAGE 41

33 expectancy with the amount of b eer consumed in the taste test the data are not conclusive evidence of a moderating effect of the prime. Therefore, the prime conditions must be examined in the context of the ERP task to determine whether or not there was an effect of the experimental manipulation. Table 4 Correlations between Drinking Variables and Beverage Consumption Beverage Consumption (ml) Drinking Variables Whole Sample (n=52) Alcohol Prime (n=24) Neutral Prime (n=28) Drinks in last month .35* .44* .33 Days Drinking in last month .35* .57* .25 Drinks / Drinking Day .25 .15 .33 Drinks per Week .37** .44* .36 Frequency .23 .53* .07 Quantity per occasion .37** .31 .42* Frequency Drunk .19 .42* .10 Frequency of beer drinking .33* .54* .28 p <.05 ** p <.01

PAGE 42

34 Table 5 Correlations between Expectancy Me asures and Beverage Consumption Beverage Consumption (ml) Expectancy Measures Whole Sample (n=52) Alcohol Prime (n=24) Neutral Prime (n=28) AEQ – Global Positive .31* .46* .22 AEQ – Social Assertion .18 .54* .06 AEQ – Aggression/Arousal .28* .50* .17 AEQ – Sexual Enhancement .11 -.10 .20 AEQ – Social & Physical Pleasure .30* .30 .30 AEQ – Tension Reduction .06 .24 -.04 AEQ – Total .27* .44* .19 AEM – Horny .29* .24 .33 AEM – Ego .11 .42* -.07 AEM – Sick .20 .14 .13 AEM – Woozy -.07 .15 -.18 AEM – Soc .05 .23 -.06 AEM – Attract .12 .05 .16 AEM – Sleep .13 -.01 .27 AEM – Danger -.00 .02 -.00 AEM – Sedating .08 .10 .08 AEM – Negative .05 .26 -.04 AEM – Positive/Arousing .20 .20 .22 p <.05 ** p <.01

PAGE 43

35 Oddball Task Averaged Waveforms The oddball task served as an index of the P300 elicited by rare events or unexpected ev ents. Infrequent Target stimul i were expected to elicit a larger positivity compared to the frequent Stan dard stimuli. ERPs were first examined in the averaged waveforms. Figure 1 depicts the averaged waveforms for a frontal electrode (Fz), a central electrode (Cz), and a parietal electr ode (Pz) for all participants comparing responses to Target and Standard stimuli. The classic P300 is typically largest over parietal electrode sites. In Figure 1, it is apparent that a widely distributed positivity around 220 ms occurs in response to the rare Ta rgets, but not to the frequent Standard stimuli. In the averaged waveforms at Pz a similar positive peak is observed around 220 ms that differentiates the stimuli, and this is followed by a second positive peak around 350-360 ms that also differentiates the stimu li with rare Targets eliciting a larger positivity than Standards. The early positivity is likely a component called the P3a, which responds indiscriminately to infrequent or nove l targets; while the later parietal positivity is most likely the P3b (or “cl assic” oddball), which is asso ciated with context updating that is pertinent to the current study (Donchin & Coles, 1988; Squires, Squires, & Hillyard, 1975; Polich, 2007). A subsequent PCA allowed for further these components to be isolated and provided values for hypothesis testing.

PAGE 44

36 Figure 1 Oddball Averaged Waveforms at Fz, Cz, & Pz electrode sites Milliseconds V +

PAGE 45

37 Spatiotemporal PCA The spatial PCA was conducted first, to reduce the dataset from 129 channels to a smaller set of “virtu al electrodes” that represent clusters of variance in the data. Fourteen spatial f actors (SFs) were rotated using Varimax (orthogonal) rotation, accounting for 87% of the variance. Of the 14 SFs, only the first few yielded interpretable spatia l topographies (Figure 2 depi cts topography of the first 5 SFs). The first spatial factor, SF1, depict s a frontal component. SF2 exhibited the strongest loading around fronto-cen tral, slightly right lateral electrodes, while SF3 loaded highly in the centro-parietal area typically associated with the P300 component. SF5 may be associated with response activation from the motor cortex as most participants were right handed and a response was required for e ach trial. The other sp atial factors did not appear to capture significant variance associat ed with the task. As demonstrated in the literature, and confirmed in this sample by examining the averaged waveforms, it was expected that the P300 would load most highl y in the centro-parietal region, so SF3 was keyed upon for further analysis thou gh other SFs were also examined. Figure 2. Spatial Factor Loadings (Virtual Electrodes) from PCA on Oddball Data

PAGE 46

38 The spatial factors provided loadings for each of the 129 electrodes, so that higher loadings are more influential on the factor scores which are derived for each of the observation points (i.e., time point s in the case of ERPs). Thus factor scores for each of the spatial factors of interest become “virtual electrodes” that were plotted over time yielding “virtual ERPs” comparable to averag ed waveforms. Figure 3 depicts the virtual ERPs for SF3, SF2, and SF1, which corres pond roughly to the averaged waveforms presented for Pz, Cz, and Fz, respectively. In both SF2 and SF3 the waveform elicited by Target stimuli exhibited a larger positive peak than that elicited by Standard stimuli in the 300-400 ms range as expected, and in SF1 Target stimuli elicited a larger positive peak in the 200 ms range (see discussion above of the P3a). The results of the spatial PCA were then submitted to a temporal PCA to identify particular areas across the 1200ms recording in which variance clustered. Ten factors accounting for 94% of the variance were agai n rotated using Varimax rotation. Figure 4 depicts the temporal factors (TFs) plotted as factors loadings ac ross time to show the factor peaks at particular time points. Se veral factors overlapped with the epoch of interest identified in the av eraged waveforms. Specifically, TF2 (peak around 400ms), TF4 (peak around 300ms), and TF6 (peak aroun d 250ms) cover the range of time in which we expected to detect the P300. By extracting the factor scores where the particular SFs and TFs of interest overlap, inferential statistics can be used to test hypotheses. It was expected that waveforms elicited by rare Target stimuli would be more positive at parietal scalp locations in the 300400 ms range. Thus, we first tested the SF3– TF2 combination and determined that the expected results were obtained [ t (54) = -7.96,

PAGE 47

39 Figure 3. Virtual Electrodes for SF1, SF2 & SF3 V + Milliseconds

PAGE 48

40 p <.001]. Thus, the Oddball task confirmed the presence of a P300 in this sample of participants that is exhibite d at parietal scalp sites betw een 350-400 ms following simple rare stimuli as compared to simple frequent stimuli. The waveforms following Target stimuli were also more positive in SF2–TF2 [ likely an index of the same phenomenon in a slightly more frontal location; t (54) = -4.84, p <.001], as well as in SF1–TF6 [i.e., the P3a; t (54) = -14.59, p <.001]. The measure of the P300, particularly the SF3–TF2 index, provides evidence that this sample of particip ants exhibits a typical P300 response in rare or unexpected stimuli, thus meeting a boundary condition for subsequent analysis of the ERPs elicited by the sentence paradigm. Figure 4. Temporal Factor Loadings from PCA on Oddball Data Factor Loadings

PAGE 49

41 Expectancy Sentence Task Averaged Waveforms ERPs were first visually examined in the averaged waveforms. This allows a cursory examina tion of the results a nd provides a frame of reference for the subsequent decomposition of the averaged waveforms. Figure 5 depicts the averaged waveforms for a frontal electr ode (Fz), a central electrode (Cz), and a parietal electrode (Pz) for al l participants and for the 4 se ntence conditions of interest: Alcohol Positive, Alcohol Negative, Incongruent, and Other. The waveforms at Fz and Cz exhibit the typical N1/P2 complex, which indexes attentional processing. Cz and Pz are then characterized by a negative defl ection in the 400-500 ms range followed by a positive deflection that continues for se veral hundred milliseconds. The sentence conditions appear to diverge particularly beyond 400 ms, which highlights an epoch of interest for additional analyses. Expectancy and priming effects Since the hypotheses of this study predict an effect of expectancy on ERP response and an interaction between prime and expectancy, the averaged waveforms were split by expectancy variables and prime for visual inspection. For the purposes of presentation, me dian splits were performed on the sample to create low and high exp ectancy groups for specific expectancy measures. The one positive and one negative expectancy measure was selected based on the correlations with variables indexing typical drinking (Table 3), because these expectancy scales best approximate the relationship between cogni tive schemas related to alcohol and the outcome behavior of interest in this sample. Therefore, median splits were derived for the AEMax Woozy scale (Figure 6), and for th e AEMax Attractive scale (Figure 7).

PAGE 50

42 Figure 5 Average Waveforms at Fz, Cz, & Pz electrode sites for Sentence Task V + Milliseconds

PAGE 51

43 Hypotheses for the High Woozy compared to the High Attractive groups would predict a different and potentially opposite effect in response to alcohol positive and negative sentence endings. Positive sentence endings were hypothesized to be more unexpected in the High Woozy group leading to a larger P300, and negative sentence endings were hypothesized to be more unexpect ed in the High Attr active group leading to a larger P300. The effect of the prime was hypothesized to increase responding in expectancy congruent ways for individuals with higher positive e xpectancies. That is, negative sentence endings were expected to elic it larger P300s in individuals with higher positive expectancies following the alcohol prime. No hypothesis was made concerning the effect of the prime on individu als with higher negative expectancies. When split by the AEMax Woozy scale (Figure 6), there was a pronounced difference between High and Low Woozy groups in the Alcohol prime condition starting at about 600 ms and persisti ng to 900 ms following negative alcohol sentences that was evident at all three el ectrode sites (Figure 6, right colu mn, dark lines). The High Woozy group exhibited a larger positivity compared to the Low Woozy group. A similar effect was evident following positive alcohol sentences, but only at the parietal electrode site (Figure 6, left column, dark lines). These di fferences were not evident in the Neutral prime group (Figure 6, grey lines); however, the High Woozy group in the Neutral prime condition did tend to exhibit a larger negativity in the 400-500 ms range at several electrode sites following both negative and pos itive alcohol sentences compared to the Low Woozy group (Figure 6, grey lines). While not hypothesized, these graphs may suggest that the alcohol prime enhanced the differences in responses to alcohol-related sentences between thes e expectancy groups.

PAGE 52

44 Figure 6. Prime groups compared between AE Max Woozy median split groups + V AlcoholPositiveSentences Alcohol Negative Sentences Milliseconds

PAGE 53

45 Figure 7. Prime groups compared between AEMax Attractive median split groups AlcoholPositiveSentences Alcohol Negative Sentences Milliseconds + V

PAGE 54

46 The AEMax Attractive median split wavefo rms (Figure 9) look rather different. One notable difference was evident following negative alcohol sent ence endings at Pz, where the Low Attractive group in the al cohol prime condition exhibited a large positivity from 500-900 ms compared to the High Attractive group (Figure 7, right column, bottom graph, dark lines). This effect is similar to that in the AEMax Woozy waveforms, in that sentence endings that were assumed to be more congruent with expectancies actually elicite d a larger positivity for the Alcohol prime condition. These figures provide only perfuncto ry evidence of the phenomena occurring in the waveforms as they are crudely produced with median split groups and do not provide a way to systematically test the hypotheses. Clearly, th ere are differences that need to be sorted out, and additional analytical tools utilized below provide a more precise test of the hypotheses. Spatiotemporal PCA The spatial PCA was conducted first again, to determine virtual electrodes. Fourteen SFs were rota ted using a Varimax rotation, accounting for 83% of the total variance. Of the 14 SFs, only the first few yielded interpretable spatial topographies (Figure 8 depicts topography of the first 6 SFs). The first spatial factor, SF1, appeared to be an artifact from eye movement around the le ft eye. Although data were processed to remove eye blinks and eye movements, the methodology is imprecise and sometimes leaves such artifacts behind. SF2 exhibited the strongest loading around electrodes that sometimes re flect the N400 component (fr onto-central, slightly right lateral), while SF3 loaded highly in the centroparietal area typically associated with the P300 component (as seen in the Oddball data above). The other spatial factors did not appear to capture significant vari ance associated with the task.

PAGE 55

47 Figure 8. Spatial Factor Loadings from PCA The virtual ERPs resulting from the SFs were then examined. As with the averaged waveforms, the sample was split along several variables of interest in order to examine the waveforms with the hypotheses in mind. Figure 9 depicts virtual ERPs SF3 with the sample split by AEMax Attractiv e, AEMax Woozy, and prime for the two alcohol sentence conditions. For SF3, there app ear to be large differences between the expectancy groups particularly in the Al cohol prime condition from 500-900 ms. In addition, in the Neutral prime group there appe ar to be larger nega tivities in the 400-500 ms range. Overall, the differences tend to a ppear in the later epoch (i.e., >400 ms). In order to determine specific temporal epochs over which differences can be tested, these data need to be run through a temporal PCA. The spatial factors were submitted to a temporal PCA to identify particular epochs across the 1100 ms recording in which varian ce clustered. Ten factors, accounting for 92% of the variance, were rotated using Varimax rotation. Figure 12 depicts the TFs plotted as factors loadings acro ss time to show the factor pe aks at particular time points. Several factors overlapped with the epochs of interest identified in the averaged waveforms. Specifically, TF1 (peak between 800-900 ms), TF3 (peak around 550 ms), and TF5 (peak around 450 ms) cover the time frames in which the average waveforms appeared to differ. Based on a prior study, the target temporal region for the P300

PAGE 56

48 Figure 9. Virtual ERPs for SF3 (centro-parietal) was between 300 and 600 ms, and the TF that was most similar to that reported in a previous study using this paradigm was TF 3 (Fishman et al., 2008). An examination of the averaged waveforms and virtual ERPs, how ever, suggested that the positive peak in this study occurred later than the epoch encompassed by TF3. Therefore, the TFs surrounding TF3 were also examined covering the epochs from 300 to 900 ms. MillisecondsAEMax Woozy Median Split AEMax Attractive Median Split

PAGE 57

49 Figure 10. Temporal Factor Loadings from PCA Analyses of SF–TF scores SF–TF scores were examined as outcome va riables in a series of regressions to test for the hypothesized eff ects of expectancy and primi ng on ERPs. Predictor variables were chosen based on the hypothesized relatio nships between expectancy and sentence types and on the correlations of the expectan cy measures with dri nking variables (Table 3). Thus, AEMax Attractive was selected as the positive expectancy scale and AEMax Woozy was selected as the ne gative expectancy scale. In addition, three higher-order expectancy scales, the AEMax Positive and Arousing scale, the AEMax Sedating scale, and the AEQ Global Positive scale were also selected, because these scales reflect a

PAGE 58

50 broader range of expectancy concepts. Prim e group and an interaction variable between expectancy and prime group were entered for each of the regressions. SF3–TF3 and SF3– TF1 were examined as potential P300 components, while SF3–TF5 was examined potentially as the N400 component (based si mply on the examination of the averaged waveforms and virtual ERPs). A total of 30 regressions were conducted; ten for each of three SF–TF combinations, using each of th e 5 expectancy variables predicting either positive or negative alcohol sentence scores. Table 6 contains the results of regressions with significant or near significant predictors. SF3–TF3 Based on previous research, the spa tio-temporal factor scores for SF3– TF3 seemed to be a good candidate fo r the P300 effect; however, none of the hypothesized predictors were significant. While the averaged waveforms and the virtual ERPs showed some differences in this time epoch, the regression analyses indicate that this SF–TF score does not react to the task pa rameters to which the P300 was predicted to respond and should not be considered the P 300 component. It appear s from the virtual ERPs that this epoch does not encompass the positive peak, which o ccurred slightly later than the peak for TF3. SF3–TF1 The second temporal dimension to be examined was the latest and the one with the widest temporal span. While this temporal factor covers the very latest time window in which the P300 is typically obser ved, the virtual ERPs (Figure 9) show positive deflections that differ by condition in th e waveforms in this later section of the epoch. The regression model using AEMax A ttractive to predict scores elicited by negative alcohol sentences was significant [Table 6; F (3,51) = 3.17, p < .05]. In this case, both Prime and the interacti on between Prime and AEMax Attractive are significant

PAGE 59

51 predictors (Table 6). Overall, individual s in the alcohol prime group exhibited more positive scores on this factor (Figure 11). The interaction in this model appears to indicate that there is a slightly positive relationship between AEMax Attractive and SF3– TF1 scores in the neutral prime condition ( r = .17; i.e., as AEMax increases, SF3–TF1 scores become more positive), but in the al cohol prime condition this relationship is negative (Figure 11; r = -.39, p = .06). The result from the Neutral prime group appears to replicate the relationship between expectancy and ERPs of the pr evious study, though the correlation is not significant. This appears to support the not ion that SF3–TF1 represents the P300 in this study, and individuals in th e neutral prime condition tend to exhibit the hypothesized reaction to the negative alcohol sentence endings with a larger positive deflection to stimuli that are less expect ed based on their self-reported alcohol expectancies. Interestingly, this relationship is reversed in the Alcohol prime condition so that the higher expectancies exhib it less positivity to these stimuli. AEMax Attractive also exhibi ted a trend toward signifi cantly predicting responses to positive alcohol sentences at this SF–TF combination. Though the total model was not significant, AEMax Attractive scor es tended to predict scores in the expected direction ( = -.24, p = .08). When the underlying correlation s were examined, it was evident that AEMax Attractive scores were negatively correlated with SF3 –TF1 scores in the Neutral prime group ( r = -.39, p <.05), but were not significantly co rrelated with these scores in the Alcohol prime group ( r = -.12). Once again, this finding seems to replicate the prior study in the Neutral prime group. In order to further examine the di fferential effects of the primes, the correlations between SF3–TF1 factors and other expectancy measures were examined by prime group. The AEMax Soci al scale also exhi bited a significant

PAGE 60

52 correlation in the expected direc tion in the Neutral prime group ( r = .36, p <.05), but not in the Alcohol prime group ( r = -.01). These results point to ward the fact that something may have been occurring between the groups that was potentially attributable to the prime condition. AEQ Global Positive scores also predicted SF3–TF1 scores following negative sentences in a similar manner as the AEMax A ttractive scores (Table 6). That is, overall the Alcohol prime group exhibited higher scores on SF3–TF1, and an interaction occurred between AEQ Global Positive and pr ime condition that led to a similar inverse relationship (Figure 12). For the Neutral prime group there wa s a trend toward a positive relationship indicating a larger P300 to negative sentences ( r = .30, p =.10), while in the Alcohol prime group the relationshi p was non-significant and negative ( r = -.25). Also similar to the results from AEMax Attractive, the AEQ Global Positive scores exhibited a relationship with SF3–TF1 scores following pos itive sentences as well. The total model was not significant, but AEQ Global Positive scores predicted SF3–TF1 scores in the expected direction indicating higher expectancy scores were related to smaller P300s following positive sentence endings ( = -.29, p < .05). The underlying correlations indicated that AEQ Global Po sitive scores were correla ted negatively with SF3–TF1 scores in both prime conditions, though thes e relationships did not reach significance. Thus, several expectancy scales provide d support for the hypothesis that the P300 responds to expectancy viola tion in the Neutral prime cond ition; however, th e results for the Alcohol prime condition were not as predicted. SF3–TF5 Temporal Factor 5 (peaking ar ound 450 ms), which immediately precedes TF3, was the last temporal dimens ion examined. Based on the virtual ERPs

PAGE 61

53 (Figure 9), this epoch appears to encompass a negative peak in the waveform. So, this SF–TF combination was not considered for the P300 but it was examined for potential differences in the range of the N400. While we did not expect differences in the N400 since this component is typi cally elicited by semantically incongruent sentences, these analyses were conducted to explore whethe r the subjective incongr uence based on one’s alcohol expectancies may aff ect this negative peak in the waveform even though the sentences were semantically congruent. One regression model significantly predicte d scores for this SF–TF combination. The results for the regression predicting SF3–T F5 scores for negative alcohol sentences using AEMax Positive Arousing are presented in Table 6 [ F (3,51) = 3.4, p < .05]. Neither AEMax Positive Arousing nor Prime significa ntly predicted scores in the model; however, there was an interacti on between expectancy and Prime. It appears that there is a negative relationship between AEMax Posi tive Arousing and SF3–T F5 scores in the alcohol prime condition (i.e., as AEMax scores go up the ERP is more negative; Figure 13), but this relationship is slightly pos itive in the neutral prime condition (i.e., as AEMax scores go up the ERP is less negative) It is unclear whether this negative deflection is the N400 component but this factor appears to be affected by priming and expectancy and it may index some cognitive process related to expectancy. In addition, since this negative deflection temporally pr ecedes the positive deflection being examined as the P300 and is affected by factor s hypothesized to affect the P300, it may significantly influence the m easurement of subsequent positivity. The prior study using this paradigm did not report differences in ne gative peaks associated with expectancies, but it appears worth considering for future studies given the resu lts obtained here.

PAGE 62

54 Table 6 Summary of Regression Analys es for Models Predicting SF – TF Factor Scores Outcome Variable Predictor Variables B SE B p SF3–TF1 Alc. Negative* AEMax Attractive .03 .03 .15 .36 Prime 1.38 .46 .92 .00 Attractive X Prime -.12 .05 -.71 .04 SF3–TF1 Alc. Negative* AEQ Global Pos .04 .03 .30 .11 Prime 1.06 .35 .71 .00 Positive X Prime -.08 .04 -.55 .05 SF3–TF5 Alc. Negative* AEMax Pos. Arousing .01 .02 .10 .57 Prime 1.42 .86 .83 .10 Pos. Arouse X Prime -.06 .03 -1.15 .03 p -value for model <.05 Figure 11 Scatterplot of SF3–TF1 scores for Negative Alcohol Sentences by Prime & AEMax Attractive

PAGE 63

55 Figure 12 Scatterplot of SF3–TF1 scores for Negative Alcohol Sentences by Prime & AEQ Global Positive Figure 13. Scatterplot of SF3–TF5 scores for Negative Alcohol Sentences by Prime & AEMax Positive Arousing ERPs and drinking variables Lastly, the SF–TF scores were compared to drinking variables to examine if there were associations between dr inking behavior (both self-reported and in-lab) a nd brainwave reactions to alcohol-related sentences. Correlations are presented in Tabl e 7. None of the factor scor es were significantly related

PAGE 64

56 to in-lab drinking. The single-item Frequenc y score was negatively correlated with ERP responses following positive alcohol sentences in the P300 (SF3–TF1). This relationship represents what was hypothesized in that pe ople who drink more and drink more often would be more likely to find positive alcohol sentences as fitting with their expected outcomes and so would exhibit a smaller posi tivity compared to people who drink less frequently and who might have less positive as sociations with alcohol. While Frequency was the only variable to exhibit a significant relationship, it provides evidence in support of the theory connecting expectancy and dr inking experience with ERP reactions to alcohol-related sentences. Table 7 Correlations between Drinking and SF3 – TF Factor Scores TF1 TF3 TF5 Alc Pos Alc Neg Alc PosAlc Neg Alc Pos Alc Neg mL beverage consumed .09 .00 .10 .05 -.11 -.14 Days Drinking in last month -.23 -.03 -.11 -.08 -.21 -.01 Drinks / Drinking Day -.07 -.03 .10 .08 -.04 .01 Drinks per Week -.20 -.07 -.04 -.06 -.10 -.04 Frequency -.29* -.12 -.08 .04 -.21 -.05 Quantity per occasion -.15 -.12 -.14 -.05 -.10 -.02 Frequency Drunk -.15 .03 -.07 .12 .05 .05 *p > .05

PAGE 65

57 Discussion ERPs elicited by sentences that violate one ’s subjectively held expectations have been shown to elicit a P300 (Fishman et al., 2008). In this study, we sought to replicate this finding and explore whether exposing indi viduals to a video prime prior to an ERP task would alter the context of alcohol rela ted sentences in such a way that the ERPs would be augmented. In the current sample we found support of the effect that was previously reported, but in a sli ghtly different instantiation. Prime – Expectancy Interactions Evident in ERPs There was some support for the hypothesis that individuals with lower positive expectancies exhibit a larger P300 to negative and sedating sentence endings compared to individuals with higher positive expectancies Similarly, there was some evidence that individuals with higher negative expectanci es exhibit a larger P300 to positive and arousing sentence endings compared to individuals with lower negative expectancies. Though these trends were evident in the wavefo rms (particularly in the virtual ERPs), the only significant correlation be tween expectancy measures a nd P300 factor scores was for the AEQ Global Positive scale relating to the SF3–TF1 scores following positive alcohol sentences. This relationship does support the hypothesis that the more positive one’s expectancies, the smaller the P300. The results point toward a slight ly more complicated picture, however, when the effect of the primes was examined. The interactions in the regression analys es indicated that the Neutral prime group exhibited the expected relationship between positive alcohol expectancies and factor scores from the negative alcohol senten ce condition, replicati ng the prior study. The Alcohol prime group, on the other hand, e xhibited higher overall P300 scores but

PAGE 66

58 displayed a negative relationship between expe ctancy and P300 amplitude. In this case, it appears that the result of the prime actuall y decreased the P300 to negative sentence endings for individuals with higher positive expectancies. In contrast, the Alcohol prime group displayed stronger correlations between several expectancy measures and ad lib drinking as well as between se lf-report drinking measures a nd ad lib drinking compared to the Neutral prime group. These behavioral results indicate an increase in the congruence of expectancies with the ad lib consumption perhaps refl ecting an increase in the activation of the individua ls’ alcohol expectancies, but the prime groups did not differ in the amount of beer they consumed in the drinking session. Taken together these trends do hint toward the possibility that the pre-exposure to an al cohol context influenced both ERPs and ad lib drinking, but the effect on ad lib drinking seems to be in an expectancy congruent way and the effect on ERPs is more difficult to determine. Thus, it appears that the Alcohol prime may have activated expectan cy networks, which altered the context in which the participants completed the tasks in the experiment, leading to ad lib drinking that was more congruent with self-reported expectancies. In the Neutral prime condition it is possible that the context of the experime nt was not altered in a way that activated alcohol expectancies sufficiently to alter subsequent drinki ng. So, for individuals in the Neutral prime condition, the sentences were ev aluated simply within the context of a computer task with less association to act ual alcohol use. Thus, responses appear to reflect the individual’s cognitive appraisal of alcohol use (i.e ., their expectancies), which resulted in the P300 effect that appe ared to replicate previous findings. The Neutral prime group exhibited some of the predicted expectancy effects, but the strength of these effects were muted comp ared to the Fishman et al. (2008) study. The

PAGE 67

59 results from the Alcohol prime group are more difficult to interpret. Presumably, even in the context of a computer task, the particip ant’s pre-existing alcohol expectancies would inform the ERP responses to alcohol-related sentences. If the current hypotheses were supported, the Alcohol prime group would have exhibited similar ERP effects as the Neutral prime group but perhaps slightly la rger in magnitude (reflecting increased activation which leads to increased violati on of the activated network). One potential factor that could have affect ed the results is that the tw o prime groups differed somewhat on some of the reported drinking variables. It appeared that the Alcohol prime group may have been more experience with alcohol overa ll since they reported more drinks in the previous month and slightly higher quanti ties and frequencies of drinking (Table 2). While these differences were not statistical ly significant nor did they manifest in differences on expectancy measures, the reported differences must be considered as they would potentially explain a muted response in the less experienced drinking group (i.e., the Neutral prime group). In addition, the cu rrent sample overall drank less than the sample reported in Fishman et al. (29.8 vs. 22.2 drinks per month), which makes it difficult to directly compare results between the two studies and ma y also explain some of the muted effect in the Neutral prime group in the current study. In other words, a lighter drinking sample may have had less late nt expectancy “energy” to activate in the experiment potentially resulting in an as ymptote level of expectancy violation. The “P300” The relationship between expectancy a nd ERPs that app ears to support the hypothesis occurs rather late in the waveform (TF1). While the temporal factor that produced the trends toward significant results was rather diffuse (loading highly over a

PAGE 68

60 300 ms span), previous studies have show n that components occurring at similar temporal offsets can include the P300 (e .g., Cacioppo et al., 1996). The phenomenon that has been called the “late positive complex” is a prolonged positive waveform usually 600-900 ms after a stimulus, which is essent ially what was observed in this study. The late positive complex has been parsed using PCA methodology and it was determined that the P300 does make up part of this “complex” (Dien, Spencer, & Donchin, 2004). Also, the latency of the P300 in more dema nding tasks like the one employed in this study may be as late as 600 ms from th e stimulus as the P300 is an endogenous component reflecting internal categorizati on and decision making and as such occurs approximately 300 ms from the in ternal categorization decision not 300 ms from the actual stimulus (Dien, Sp encer, & Donchin, 2004). Another factor that could be influencing the ERPs is what is termed sentence wrap-up effects (Hagoort, 2003). These effects are observed when ERPs are recorded to the final word in a sentence and are thought to occur due to an overlap of local (word level) and global (sentence level) processing that culminates in a decision or categorization at the end of a sentence. They may account somewhat for the negative peak observed at central and parietal elec trode sites between 300-500 ms as well as the lack of a well-defined positive peak in the 500-900 ms range (Hagoort, 2003). Sentence wrap-up effects do not eliminat e the P300, but they may attenuate it or cause a flattening of the wave due to an increase in latency jitter since the senten ce level processing may occur at slightly different rates across part icipants. Thus, the positivity observed between 500-900 ms in the parietal region of most of the waveforms is likely the P300, but it is

PAGE 69

61 probably obscured somewhat do to the other fa ctors inherent to the sentence presentation paradigm. The P300 and Decisions to Use Alcohol The results of this study lend additional support to th e theory that the P300 indexes violations of subjective expectati on. The P300 response di d not correlate with subsequent consumption in the ad lib drinki ng portion of the study, but it did correlate with some of the self-reported drinking vari ables. In addition, prim ing with an alcohol cue prior to the ERP task seemed to have some effect on the observed brainwave reaction to the alcohol related sentences but it too did not appear to influence the amount of beer consumed in the ad lib drinking session direc tly. The tests of inter actions between prime condition and expectancies or drinking were not highly powered enough to detect differences, but it did appear th at the prime led to behavior that was more congruent with expectancies and drinking history. Thus, th e results of the current study still leave many questions unanswered concerning the role of the P300 in deci sions to use alcohol. It is possible that the P300 serves as a red flag to some extent that signals a violation of expectancy requiring the individu al’s current model to be alte red to some extent given the new information (i.e., the context updating hypothesis). This study would fit with the context updating hypothesis of the P300, but it does not provide additional information as to whether the P300 feeds into the decision making process that leads to subsequent drinking. It is well established that self-reported alcohol expectancies predict subsequent drinking, and activating alcohol expectancy semantic networks can lead to increases in expectancy congruent behavi or (Reohrich & Goldman, 1995; Carter et al., 1998). One of

PAGE 70

62 the aims of this study was to attempt to identify the P300 in the stream of decision making from expectancy activation via the vide o prime to drinking behavior in the taste test drinking session. Since the study fell short of this goal, future st udies should continue to explore the potential role this phenomenon plays in deci sion making. While it appears that ERPs occurring within a few hundred milliseconds after seeing a stimulus index violations of alcohol expectancies in some cases, it is still unknown how this ERP response affects subsequent behavioral decisions It is possible that such effects are more specific than the current experiment allowe d. For example, some expectancy scales exhibit stronger correlations with drinking than other, and these relationships differ by sample. It is possible that ER Ps reflect individual differences to such an extent that responses to particular words that are salien t for an individual would be more indicative of the individual’s expectancies than the expe ctancy scale scores which average together responses to several related words. Future st udies could examine responses to individual words and their relationship to subsequent decisions to drink. A full examination of the phenomenon woul d need to incorporate the underlying brain function that is putative ly related to the P300 to bett er understand the role of the P300 and how it relates to systems associated with decision making in the context of appetitive rewards. For example, the P300 has been associated with the locus coeruleus– norepinepherine system (LC–NE), which appear s to play a role in stimulus evaluation and decision-making (i.e., potentiating a response to a motivationally significant stimulus; Nieuwenhuis, Aston-Jones, & C ohen, 2005). These potentiated responses may indicate that evaluati on of motivationally significant stimu li (i.e., drinking related words, pictures, environments) is accomplished quick ly through the function of the LC–NE, and

PAGE 71

63 it may implicate this system in the initiation of decision response patt erns that eventually lead to drinking behavior. Cl early, the results of the curr ent study not provide specific evidence of such brain function, but they coul d provide the basis for future studies to examine the underlying neural systems (e.g., Polich & Criado, 2006). Conclusion The current study supports the theory th at the P300 reflects violations of subjective expectancy. This ERP effect was observed following a non-alcohol context prime, but in individuals who were exposed to an alcohol context prime the results were less clear. Specifically, P300 responses elicited by negativ e and sedating sentence endings were positively correlated with pos itive alcohol expectan cy measures (AEQ Global Positive and AEMax Attractive) in i ndividuals who viewed the non-alcohol prime and not in individuals who viewed the alc ohol prime. Thus, viewing the alcohol prime appeared to change the relationship be tween self-reported ex pectancies and ERP responses in a way that was not expected. Rather than enhancing the incongruence of negative words completing sentences about al cohol for individuals with higher positive expectancies, this effect actually appeared to mute or even reverse the expected relationship. The limitations in sampling a nd experimental design may have precluded the study from addressing the larger questi on of the role the P300 response plays in decisions to use alcohol, but the current results encourage the continued exploration of this phenomenon.

PAGE 72

64 References Babor, T. F., Brown, J., & del Boca, F. K. (1990). Validity of self-reports in applied research on addictive behavi ors: Fact or fiction? Behavioral Assessment, 12 (1), 531. Bargh, J. A., & Williams, E. L. (2006). The Automaticity of Social Life. Current Directions in Psychological Science, 15 (1), 1-4. Bartholow, B. D., Henry, E. A., & Lust, S. A. (2007). Effects of alcohol sensitivity on P3 event-related potential r eactivity to alcohol cues. Psychology of Addictive Behaviors, 21 (4), 555-563. Begleiter, H., Porjesz, B., & Bihari, B. (1987) Auditory brainstem potentials in sons of alcoholic fathers. Alcoholism: Clinical and Ex perimental Research, 11 (5), 477480. Begleiter, H., Porjesz, B., Bihari, B., & Ki ssin, B. (1984). Event-re lated brain potentials in boys at risk for alcoholism. Science, 225 (4669), 1493-1496. Brandon, T. H., & Baker, T. B. (1991). The Smoking Consequences Questionnaire: The subjective expected utility of smoking in college students. Psychological Assessment, 3 (3), 484-491. Brown, S. A., Christiansen, B. A., & Gold man, M. S. (1987). The Alcohol Expectancy Questionnaire: An instrument for the a ssessment of adolescent and adult alcohol expectancies. Journal of Studies on Alcohol, 48 (5), 483-491. Brown, S. A., Goldman, M. S., Inn, A., & Anderson, L. R. (1980). Expectations of reinforcement from alcohol: Their doma in and relation to drinking patterns. Journal of Consulting and Clinical Psychology, 48 (4), 419-426.

PAGE 73

65 Cacioppo, J. T., Crites, S. L., Jr., & Gardne r, W. L. (1996). Attitudes to the right: Evaluative processing is associated with lateralized late positive event-related brain potentials. Personality and Social Psychology Bulletin, 22 (12), 1205-1219. Carter, J. A., McNair, L. D ., Corbin, W. R., & Black, D. H. (1998). Effects of priming positive and negative outcomes on drinking responses. Experimental and Clinical Psychopharmacology, 6 (4), 399-405. Cuthbert, B. N., Schupp, H. T., Bradley, M. M., Birbaumer, N., & Lang, P. J. (2000). Brain potentials in affect ive picture processing: Covariation with autonomic arousal and affective report. Biological Psychology, 52 (2), 95-111. Darkes, J., & Goldman, M. S. (1993). Exp ectancy challenge and drinking reduction: Experimental evidence for a mediational process. Journal of Consulting and Clinical Psychology, 61 (2), 344-353. Darkes, J., Greenbaum, P. E., & Goldman, M. S. (2004). Alcohol Expectancy Mediation of Biopsychosocial Risk: Comp lex Patterns of Mediation. Experimental and Clinical Psychopharmacology, 12 (1), 27-38. Del Boca, F. K., & Noll, J. A. (2000). Truth or consequences: The validity of self-report data in health services research on addictions. Addiction, 95 (Suppl3), S347-S360. Dien, J., Spencer, K. M., & Donchin, E. (2003). Localization of the event-related potential novelty response as defi ned by principal components analysis. Cognitive Brain Research, 17 (3), 637-650. Dien, J., Spencer, K. M., & Donchin, E. ( 2004). Parsing the late positive complex: Mental chronometry and the ERP compone nts that inhabit the neighborhood of the P300. Psychophysiology, 41, 665-678.

PAGE 74

66 Donchin, E. (1969). Discriminant analysis in average evoked response studies: The study of single trial data. Electroencephalography & Cl inical Neurophysiology 27 (3), 311-314. Donchin, E. (1981). Surprise!... Surprise? Psychophysiology, 18 (5), 493-513. Donchin, E., Callaway, E., Cooper, R., Desmedt, J. E., Goff, W. R ., Hillyard, S. A., & Sutton, S. (1977). Publication criteria fo r studies of evoked potentials (EP) in man. In J. E. Desmedt (Ed.), Progress in clinical neurophysiology: Vol. 1. Attention, voluntary contraction and event-related cerebral potentials (pp. 1-11). Basel: Karger. Donchin, E., & Coles, M. G. (1988). Is the P300 component a manifestation of context updating? Behavioral and Brain Sciences, 11 (3), 357-427. Duncan-Johnson, C. C., & Donchin, E. (1977). On quantifying surprise: The variation of event-related potentials w ith subjective probability. Psychophysiology, 14 (5), 456-467. Dunn, M. E., & Goldman, M. S. (1998). Age a nd drinking-related differences in the memory organization of alcohol expectan ces in 3rd-, 6th-, 9th-, and 12th-grade children. Journal of Consulting an d Clinical Psychology, 66 (3), 579-585. Fabiani, M., Gratton, G., & Coles, M. G. (2000). Event-related br ain potentials. In Handbook of psychophysiology (2nd ed.) (pp. 53-84). New York, NY: Cambridge University Press. Fabiani, M., Gratton, G., Karis, D., & Donc hin, E. (1987). Definition, identification, and reliability of measurement of the P300 component of the event-related brain

PAGE 75

67 potential. In P.K. Ackles, J.R. Jennings, & M.G.H. Coles (Eds.), Advances in Psychophysiology Vol. 2. (pp. 1-78). Greenwich, CT: JAI Press. Fishman, I., Goldman, M.S., & Donchin, E. (2008). The P300 as a electrophysiological probe of alcohol expectancy. Experimental and Clinical Psychopharmacology, 16 341-356. Franken, I. H., Stam, C. J., Hendriks, V. M., & van den Brink, W. (2003). Neurophysiological evidence for abnormal cognitive processing of drug cues in heroin dependence. Psychopharmacology, 170 (2), 205-212. Goldman, M. S. (2002). Expectancy and risk for alcoholism: The unfortunate exploitation of a fundamental characteristic of neurobehavioral adaptation. Alcoholism: Clinical and Experimental Research, 26 (5), 737-746. Goldman, M. S., & Darkes, J. (2004). Alc ohol Expectancy Multiaxial Assessment: A Memory Network-Based Approach. Psychological Assessment, 16 (1), 4-15. Goldman, M. S., Darkes, J., & Del Boca, F. K. (1999). Expectancy mediation of biopsychosocial risk for alc ohol use and alcoholism. In How expectancies shape experience (pp. 233-262). Washington, DC: Americ an Psychological Association. Goldman, M. S., Darkes, J., Reich, R. R., & Brandon, K. O. (2006). From DNA to conscious thought: The influence of antic ipatory processes on human alcohol consumption. In Cognition and addiction (pp. 147-184). New York, NY: Oxford University Press. Goldman, M. S., Greenbaum, P. E., & Dark es, J. (1997). A confirmatory test of hierarchical expectancy structure and pr edictive power: Discriminant validation of the Alcohol Expectancy Questionnaire. Psychological Assessment, 9 (2), 145-157.

PAGE 76

68 Goldman, M. S., Reich, R. R., & Darkes, J. (2006). Expectancy as a Unifying Construct in Alcohol-Related Cognition. In Handbook of implicit cognition and addiction (pp. 105-119). Thousand Oaks, CA : Sage Publications, Inc. Gratton, G., Coles, M. G., & Donchin, E. ( 1983). A new method for off-line removal of ocular artifact. Electroencephalography & C linical Neurophysiology, 55 (4), 468484. Hagoort, P. (2003). Interplay between S yntax and Semantics during Sentence Comprehension: ERP Effects of Combini ng Syntactic and Semantic Violations. Journal of Cognitive Neuroscience, 15, 883-899. Hansenne, M., Olin, C., Pinto, E., Pitchot W., & Ansseau, M. (2003). Event-Related Potentials to Emotional and Neutral Stimuli in Alcoholism. Neuropsychobiology, 48 (2), 77-81. Ito, T. A., & Cacioppo, J. T. (2007). Attitudes as Mental and Neural St ates of Readiness: Using Physiological Measures to Study Implicit Attitudes. In B. Wittenbrink & N. Schwarz (Eds.), Implicit measures of attitudes (pp. 125-158). New York, NY: Guilford Press. Key, A. P., Dove, G. O., & Maguire, M. J. ( 2005). Linking Brainwaves to the Brain: An ERP Primer. Developmental Neuropsychology, 27 (2), 183-215. Kramer, D. A., & Goldman, M. S. (2003). Usin g a modified Stroop task to implicitly discern the cognitive organizati on of alcohol expectancies. Journal of Abnormal Psychology, 112 (1), 171-175.

PAGE 77

69 Kramer, D. A., & Goldman, M. S. (2003). Usin g a modified Stroop task to implicitly discern the cognitive organizati on of alcohol expectancies. Journal of Abnormal Psychology, 112 (1), 171-175. Leigh, B. C., & Stacy, A. W. (1998). Individu al differences in memory associations involving the positive and negativ e outcomes of alcohol use. Psychology of Addictive Behaviors, 12 (1), 39-46. Lust, S. A., & Bartholow, B. D. (2009). Se lf-reported and P3 event-related potential evaluations of condoms: Does wh at we say match how we feel? Psychophysiology, 46 (2), 420-424. Marlatt, G., Demming, B., & Reid, J. B. (1973) Loss of control drinking in alcoholics: An experimental analogue. Journal of Abnormal Psychology, 81 (3), 233-241. Morris, J. P., Squires, N. K., Taber, C. S., & Lodge, M. (2003). Activation of Political Attitudes: A Psychophysiological Examination of the Hot Cognition Hypothesis. Political Psychology, 24 (4), 727-745. Nieuwenhuis, S., Aston-Jones, G., & Cohen, J. D. (2005). Decision Making, the P3, and the Locus Coeruleus--Norepinephrine System. Psychological Bulletin, 131 (4), 510-532. Osterhout, L., Bersick, M., & McLaughlin, J. (1 997). Brain potentials re flect violations of gender stereotypes. Memory & Cognition, 25 (3), 273-285. Patton, J. H., Stanford, M. S., & Barratt, E. S. (1995). Factor structure of the Barratt Impulsiveness Scale. Journal of Clinical Psychology, 51 (6), 768-774. Picton, T. W. (1992). The P300 wave of the human event-related potential. Journal of Clinical Neurophysiology, 9 (4), 456–479.

PAGE 78

70 Polich, J. (2007). Updating P300: An integrative theory of P3a and P3b. Clinical Neurophysiology, 118 2128-214 Polich, J., & Criado, J. R. (2006). Neurops ychology and neuropharmacology of P3a and P3b. International Journal of Psychophysiology, 60 (2), 172-185. Porjesz, B., Rangaswamy, M., Kamarajan, C., Jones, K. A., Padmanabhapillai, A., & Begleiter, H. (2005). The utility of ne urophysiological markers in the study of alcoholism. Clinical Neurophysiology, 116 (5), 993-1018. Rather, B.C., Goldman, M.S., Roehrich, L., & Brannick, M. (1992). Empirical modeling of an alcohol expectancy memory network using multidimensional scaling. Journal of Abnormal Psychology, 101 (1), 174-183. Rather, B. C., & Goldman, M. S. (1994). Dr inking-related differences in the memory organization of alcohol expectancies. Experimental and Clinical Psychopharmacology, 2 (2), 167-183. Reich, R. R., Below, M. C., & Goldman, M. S. (2010). Explicit and Implicit Measures of Expectancy and Related Alcohol C ognitions: A Meta-Analytic Comparison. Psychology of Addicti ve Behavior, 24, 100-110. Reich, R. R., & Goldman, M. S. (2005). Exploring the alcohol expectancy memory network: The utility of free associates. Psychology of Addictive Behaviors, 19 (3), 317-325. Reich, R. R., Noll, J. A., & Goldman, M. S. (2005). Cue Patterns and Alcohol Expectancies: How Slight Differences in Stimuli Can Measurably Change Cognition. Experimental and Clinic al Psychopharmacology, 13 (1), 65-71.

PAGE 79

71 Roehrich, L., & Goldman, M. S. (1995). Implic it priming of alcohol expectancy memory processes and subsequent drinking behavior. Experimental and Clinical Psychopharmacology, 3 (4), 402-410. Schuckit, M. A. (1992). Advances in understand ing the vulnerability to alcoholism. In C. P. O'Brien & J. H. Jaffe (Eds.), Addictive states (pp. 93-108). New York, NY: Raven Press. Sher, K. J., Grekin, E. R., & Williams, N. A. (2005). The Development Of Alcohol Use Disorders. Annual Review of Clinical Psychology, 1 (1), 493-523. Slotnick, S. D. (2004). Sour ce localization of ERP genera tors. In T. C. Handy (Ed.), Event-Related Potentials: A Methods Handbook (pp. 149-166). Cambridge: The MIT Press. Sobell, L. C., & Sobell, M. B. (1992). Time line follow-back: A technique for assessing self-reported alcohol consumption. In R. Z. Litten & J. P. Allen (Eds.), Measuring alcohol consumption: Psychos ocial and biochemical methods (pp. 41-72). Totowa, NJ: Humana Press. Spencer, K. D., Dien, J., & Donchin, E. (2001). Spatiotemporal analysis of the late ERP responses to deviant stimuli. Psychophysiology, 38 (2), 343-358. Stein, K. D., Goldman, M. S., & Del Boca, F. K. (2000). The influence of alcohol expectancy priming and mood manipulati on on subsequent alcohol consumption. Journal of Abnormal Psychology, 109 (1), 106-115. Squires, N.K., Squires, K.C., & Hillyard, S.A. (1975). Two varieties of long-latency positive waves evoked by unpredictable auditory stimuli in man. Electroencephalography and Clinical Neurophysiology, 38 387-401.

PAGE 80

72 Vohs, K. D., Mead, N. L., & Goode, M. R. (2006). The Psychological Consequences of Money. Science, 314 (5802), 1154-1156. Wiers, R. W., Stacy, A. W., Ames, S. L., No ll, J. A., Sayette, M. A., Zack, M., et al. (2002). Implicit and explicit alcohol-related cognitions. Alcoholism: Clinical and Experimental Research, 26 (1), 129-137.