The Influence of Classical Conditioning Procedures on Subsequent Attention … · 2017-05-05 · cessful classical conditioning, assisting the subject in learning an association between

The Influence of Classical ConditioningProcedures on Subsequent Attention to theConditioned Brand

CHRIS JANISZEWSKILUK WARLOP*

Three experiments are used to investigate the influence of conditioning procedureson attention to a conditioned stimulus. In experiment 1. scenes presented in asequence that is consistent with prescribed conditioning procedures are shown toencourage attention to the advertised brands in subsequent product displays, Ex-penment 2 suggests that differential attention to conditioned brands can be attributedto the signaling properties the brand acquires as a consequence of conditioning.Evidence from a third experiment raises the possibility that semantic conditioningmay be responsible for the effects observed in experiments 1 and 2. The findingssuggest that current prescriptions on the use of conditioning procedures may needto be updated.

C onsumer research involving classieal conditioningtechniques has focused primarily on the transfer

of affective responses {Allen and Janiszewski 1989; Al-len and Madden 1985; Bierley, McSweeney, and Van-nieuwkerk 1985; Gorn 1982; Shimp. Stuart, and Engle1991; Stuart, Shimp. and Engle 1987). The emphasison affective responses can be attributed to the cognitiveorientation of consumer research and the dominanceofthe attitude construct vyjthin the discipline (Shimp1991). The emphasis on the transfer of responses canbe attributed to the tendency of consumer researchersto view conditioning as a learning process, similar tothat proposed by Pavlov (1927).

It is quite natural that consumer research should fo-cus on affective learning, given the historic interest thefield has exhibited in attitude formation and change.However, interest in affective responses comes at theexpense of other responses that are of comparable im-portance to the understanding of consumer behavior.Particularly relevant is the possibility that conditioningprocedures can influence attention and approach be-

'Chris Janiszewski is associate professor of marketing and LukWjriop is a graduate student at the College of Business, Universityof Florida. Gainesville, FL 32611. The authors would like to thankJoseph Alba. Chris Allen. Joel Cohen. Richard Lutz. Alan Sawyer.iind the three reviewers for their helpful comments on ihe manuscript.I he authors would also like to thank the William Cook Agency and•\nlhony Bouton for their assistance in stimulus development and[)erek Abraham. Dasha Baez. Sharon Brackett. Cheryl Cox. Susan(raft. Lori Giardino. Laura Jacobson. John Kubs. Michael Leone,Joel McCali. Andrew Mirman. Tim O'Sullivan. Gladys Reyna. Ste-phen Stark. Thomas Thurston. and Karen Yureka for their assistancein eye-tracking data collection and analysis.

171

havior. As first shown by Brown and Jenkins (1968),successful conditioning can be accompanied by a sec-ondary behavior that consists ofthe subject approachingthe conditioned stimulus. Similarly. McSweeney andBierley (1984) conclude that "a US [unconditionedstimulus] that evokes approach need not be found be-fore classical conditioning can be used to move peoplecloser to displays in supermarkets. Any reinforcer maybe used as the US to condition this (approach) behavior"(p. 621; emphasis added). Thus, the desire to approacha conditioned stimulus may be a natural by-product ofthe conditioning procedure.

To appreciate how conditioning might encourage at-tention, approach behavior, and an affective responsesimultaneously, one must view conditioning as a pro-cedure, not a proeess. When viewed as a process, con-ditioning is theoretically constrained to explaining alllearning as a response transfer between the uncondi-tioned stimulus and the conditioned stimulus (Pavlov1927). Thus, when multiple responses result from con-ditioning there must be multiple transfers, a predictionthat is at odds with Pavlovian explanations of condi-tioning processes.

In contrast, viewing conditioning as a procedure pro-vides the flexibility to consider the impact of condi-tioning on multiple learning systems. A conditioningprocedure can encourage perceptual, attentional, andconceptual systems to inductively learn about covari-ations in the environment. At the most basic level, aconditioning procedure can help the perceptual systemisolate potentially important event covariations. Oncehypotheses about the paired events have been generated,

© 1993 by JOURNAL OF CONSUMER RESEARCH, (nc. • Vol. 20 • Sepicmbcr 1993

All rights reserved. OO93-5301/94/2OO2-OO01$2.O0

172 JOURNAL OF CONSUMER RESEARCH

the perceptual system can encourage higher-orderlearning centers to allocate attention to informationabout these events. There can be implicit learning ofthe association between a conditioned stimulus and anunconditioned stimulus, and, with further testing, aconscious awareness of the contingency between theevents. Thus, covariation assessment, attention, implicitassociative learning, and contingency awareness becomenatural by-products of learning via a conditioning pro-cedure. Conditioning procedures can be used to en-courage consumers to implicitly learn about covaria-tions between the brand and meaningful events, toestablish or reinforce associations between the brandand significant attributes, and to establish associationsbetween a brand and affectively charged events.

We begin by investigating the influence of condi-tioning procedures on the allocation of attention. Inthe first section, we will review evidence on the potentialfor conditioning procedures to influence attention to aconditioned stimulus and the reasons contingencyawareness may not be necessary for this effect. Exper-iment I will demonstrate that conditioning procedurescan be used to encourage attention to a conditionedbrand, but that this effect does not seem to depend oncontingency awareness. Experiment 2 will diflêrentiatebetween two possible explanations for the influence ofconditioning procedures on the allocation of attention,concluding that subjects attend to the conditionedstimulus because it is potentially informative (Brownand Jenkins 1968; Hall and Channell 1985). Experiment3 will demonstrate that conditioning procedures can beused to reinforce associations between a conditionedand an unconditioned stimulus. The final section willdiscuss the implications ofthe conditioning procedureon the allocation of attention and associative learning.

BACKGROUND

Conditioning and Attention

Evidence for a relationship between conditioning andattention was originally provided by investigations intothe orienting response (see Lynn [1966] and Razran[ 1961 ] for review). An orienting response is the normalbodily reaction to a novel stimulus, or to a change in astimulus, and consists of changes in visceral, somatic,cognitive, and neural systems (Sokolov 1963). Fully de-veloped orienting responses, such as those observed inhumans, often involve a "specific molar reaction of'turning' toward the source ofthe stimulation" (Razran1961. p. 1 13;Spinksand Siddle 1983).

Early researchers experienced a difficult time de-scribing the role ofthe orienting response in condition-ing because the orienting response to the conditionedstimulus (CS) and to the unconditioned stimulus (US)closely resembled the unconditioned responses (UR)researchers hoped to condition (e.g., electrodermal re-sponses, heart rate changes, vasoconstriction; Razran

1961). This difficulty led some researchers to concludethat an orienting response to the US was in fact beingdirectly transferred to the CS (Razran 1961). Concur-rently, however, others concluded that the orienting re-sponse to a CS was a necessary precondition to suc-cessful classical conditioning, assisting the subject inlearning an association between the CS and US (Vin-ogradova 1959).

Subsequent investigations into the role of the ori-enting response in conditioning has continued to em-phasize its impact on learning during conditioningtrials, but the definition ofthe orienting response hasexpanded to include voluntary responses (e.g., atten-tion, approach, and contact). Whereas Sokolov (1963)saw the orienting response as a general readiness to re-spond, later theorists saw the orienting response as amechanism that directed attention to arousing stimuli(e.g.. the CS and US). For example, Ohman (1979)viewed the orienting response as a "call for processingcapacity in a central, limited capacity channel" (p. 444).According to his model, either of two characteristics ofa stimulus could be identified by the preattentive pro-cessing mechanisms that are responsible for directingattention. First, attention could be allocated when therewas a mismatch between a stimulus and expectationscreated by working memory. Second, attention couldbe allocated when there was a match between a stimulusand a preattentively primed memory representation thatwas deemed significant.

Ohman's theory (1979) is instrumental to current ex-planations ofthe role ofthe orienting response in learn-ing and conditioned responding. Pearceand Hall (1992)review animal evidence and conclude that conditioningprocedures are most effective when a subject attends(orients) to the CS and the US during initial pairings(Kaye and Pearce 1984). During the course of condi-tioning trials, the orienting response to the CS will de-cline as the subject learns that the CS predicts the US,since further attention to the CS is not relevant oncethe association has been learned (Kaye and Pearce1984). When the CS is presented in a new context, theCS suddenly takes on renewed significance (Pearce andHall 1992). The subject will once again orient to theCS. since there is a need to determine whether the CSwill continue to predict the US in the novel context(Hall and Channell 1985). In other words, the CS isinteresting because it is potentially relevant to thelearning ofthe CS-US association. If we generalize tothe consumer domain, these findings suggest that an adthat more successfully associates a CS and a US mayencourage attention (orienting) to the brand at the pointof purchase (i.e., in a novel context) because this atten-tion may further assist in the learning of the CS-USassociation.

Contingency Awareness and AssociativeLearning

The hypothesis that subjects will allocate attentionto a CS in novel contexts is intriguing but troubling.

MULTIPLE BENEFIT CONDITIONING PROCEDURE 173

given our current understanding ofthe role of awarenessin human associative learning. In studies investigatingaffective responses it has been found that contingencyawareness, awareness of the CS-US pairing, often ac-companies associative learning (Allen and Janiszewski1989: Shimp et al. 1991b). Subjects use awareness ofthe CS-US pairing to respond to the CS during test trials(i.e.. in their attitudes expressed on an affective-responsescale). Yet. once subjects are aware of the CS-US as-sociation, they should no longer engage in attempts toconfirm the association. In other words, once contin-gency awareness accompanies associative learning, thereshould not be increased attention to the CS in a sub-sequent context because the association has alreadybeen learned. Therefore, increased attention to a CS ina novel context should occur only when prior CS-USpairing suggests an association but the association hasnot yet become strong enough to result in contingencyawareness. In effect, attention to the CS in a novel con-text would be a form of hypothesis testing, a testing ofan association that is not yet strong enough to havecreated contingency awareness.

The hypothesis that associative learning in the ab-sence of contingency awareness will promote attentionto the CS in a novel context is compatible with contem-porary explanations of associative learning systems.Associative learning theorists propose a two-stagelearning system in which a perceptual system selectsfeatures and feature combinations that can be investi-gated and subsequently incorporated into rules in a sec-ond, higher-order system (Hall 1991; Holyoak, Koh,and Nisbett 1989). The perceptual system is noncons-cious. parallel, and has the ability to select from largearrays of data, identifying potential covariations forsubsequent learning by the higher-order system (Ho-lyoak etal. 1989). In effect, the perceptual system is aninformation-generation device that selects and createsthe small array of information the higher-order systemwill process. As a person learns a CS-US association,the perceptual system will gather information on thesestimuli to assist the higher-order system in assessing thestrength of this relationship. As the higher-order systemdevotes resources to assessing the strength of this rela-tionship, confirmations will often be accompanied bycontingency awareness.

The goal of this article is not to explicate the role ofcontingency awareness in an associative learning pro-cess, although the results will indirectly comment onthis issue. Instead, we wish to demonstrate that a con-ditioning procedure can promote associative learningand that a consequence of the process of learning isincreased attention to the CS. Attention to the CSshould be particularly strong when prior pairings oftheCS and US have encouraged the formation of an as-sociation hypothesis but pairings have been insufficientto allow for contingency awareness. In experiment 1, aconditioning procedure is shown to encourage attention

to a CS in a novel context, but this effect occurs in theabsence of contingency awareness.

EXPERIMENT I

Experiment 1 used television commercials to inves-tigate whether the allocation of attention to brandchoices could be influenced by a conditioning proce-dure. We expected that commercials organized to beconsistent with a forward conditioning sequence wouldencourage consumers to attend to a brand sooner thancommercials organized with a random conditioning se-quence. Attention to the brands would be assessed inthe context of a product display.

Stimuli

The stimuli were constructed with two 30-secondtelevision commercials, a Mountain Dew commercialfeaturing white-water surfing and a Canada Dry com-mercial featuring scenes of couples having fun. Theoriginal version of each commercial consisted of a se-quence of approximately 18 segments arranged tocatchy jingles. For each commercial, the 18 segmentswere divided into three groups; six segments showingthe product only (CS). six segments that were interestingor fun (US), and six segments ofthe product being con-sumed (filler segments). The CS segments prominentlydisplayed the product package and lasted from one totwo and one-half seconds. The US segments were fourto eight seconds long, often consisting of multiple cutsof a similar scene. The filler segments showed actorsconsuming or enjoying the product and often includeda nonpromincnt display of part or all of the productpackage. The filler segments lasted between one andthree seconds. These segment lengths were dictated bythe composition ofthe original ads.

Two experimental commercials were made from eachoriginal. The forward conditioning version of eachcommercial consisted of six trials. Each trial consistedof a product segment (CS). followed by an entertainingsegment (US), followed by a product-consumption seg-ment (filler). Figure 1 shows scenes from the six trials(18 segments) ofthe Mountain Dew commercial. Trial1 begins with a segment showing ice hitting a bottle ofMountain Dew (CS segment), followed by a segmentshowing teenagersjumping into a mountain stream {USsegment), and ending with a segment showing a womanconsuming a bottle of Mountain Dew (filler segment).Trials 2-6 followed an identical pattern. Each segmentbegan immediately after the other as did each trial (e.g.,a filler segment was immediately followed by a CSproduct segment).

The random conditioning versions of the experi-mental commercials consisted of a random ordering ofthe six trials representing the six possible orders of aCS, a US, and hller segments (e.g.. CS -^ US -^ filler;CS -*• filler -<• US; US - • CS - • filler; US -^ filler -^

174

175


CS: filler -» CS — US; filler -^ US -»- CS). The lowerright-hand corner of each scene in Figure 1 includesthe key word "Random" and a number indicating theorder in which this scene appeared in the random con-ditioning version of the commercial.

Design

The experiment was a between-subjects manipulationof the conditioning procedure {forward, random) witha stimulus replication.' One treatment group received18 forward conditioning trials for brand A (MountainDew) and 18 random conditioning trials for brand B(Canada Dry). A second treatment group received 18forward conditioning trials for brand B (Canada Dry)and 18 random conditioning trials for brand A {Moun-tain Dew). The conditioning trials for both groups wereaccomplished by embedding three presentations of for-ward and random conditioning commercials within asequence of six filler ads. Filler ads were all for mod-erately known brands to allow for the observation of aconditioning influence {Shimp et al. 1991b).

ProcedureSubjects. Subjects were 54 undergraduate juniors

and seniors enrolled in the senior author's consumerbehavior course {28 women. 26 men). Subjects partic-ipated in the experiment eight weeks prior to the classdiscussion of the conditioning literature.

Stimulus Presentation. Subjects were invited intothe laboratory one at a time. Upon arriving, the subjectwas placed in a private room containing a televisionand VCR and given the following instruction:

In this first experiment you will watch a series of /2 com-mercials. Ail of the commercials are for soft drinks. Someof the commercials will be repeated because you oftensee a commercial more than once during a session oftelevision viewing. Please pay careful attention to thecommercials. The experimenter will stop the tape whenit is done.

The experimenter then started the tape and left theroom.

The experimental commercials were presented as partof a series of commercials for eight different soft drinks.Each of the two experimental tapes consisted of 12commercials, each commercial separated by five sec-onds of black space. The Canada Dry commercial oc-cupied the second, sixth, and tenth positions in the se-quence of ads. The Mountain Dew commercialoccupied the fourth, eighth, and twelfth positions inthe sequence of ads. The other commercials were forMinute Maid {first position), Sunkist {third). Slice

'A random control group was not used in this design because ofresource constraints associated with the eye-tracking procedure. Arandom control group was used in experiments 2 and 3.

{fifth), Schweppes ginger ale {seventh). Orange Crush{ninth), and Seven-Up {eleventh). The filler commer-cials were selected to be inconsistent with conditioningprocedures.

Dependent-Measure Stimuli. Dependent-measurestimuli consisted of 12 computer-generated video slides,each consisting of a diamond-shaped arrangement offour of the eight brands. Each soft drink container wasdigitized with a color scanner, touched up with a high-resolution paint package, and then arranged into a videopresentation slide with Microsoft Corporation's Pow-erpoint presentation package. The containers for thetarget brands {i.e.. Mountain Dew and Canada Dry)appeared on the third, sixth, ninth, and twelfth stimulusslides, hereafter called the first, second, third, and fourthtest presentation. The test stimuli were presented witha 20-inch Multisync 5D monitor equipped with a VideoSeven high-resolution graphics card {256 colors at 800X 600 resolution).

The test presentations of interest were the second andthird in the sequence. Limiting the emphasis to the sec-ond and third presentation was a consequence of thecomputer presentation software and the characteristicsof the conditioned stimuli. First, to present the stimuliat a resolution and a color that made them look realistic,a commercial computer presentation package had tobe used {Powerpoint). Like most computer presentationsoftware, PowerPoint writes to the video screen line-by-line, beginning with the upper left-hand corner andending with the lower right-hand corner. The roll-downeffect on the screen was noticeable and had the potentialto influence attention {all pretest subjects mentionedthis problem). Second, the Mountain Dew ad used abottle as a CS, whereas the Canada Dry commercialand all the filler ads but one used cans as the picturedbrands. Pretesting demonstrated that subjects werehighly sensitive to the nonconforming shape of the bot-tle, even though it had been sized to the height of thecans {all pretest subjects stated that the bottle influencedtheir attention because it was different). To reduce theshape bias, the first test presentation consisted of a dis-play with the Mountain Dew bottle at the top of thescene and the Canada Dry can at the bottom of thescene {Seven-Up on the lefit, Schweppes on the right).

The second presentation had the Mountain Dew bot-tle placed on the left point of the diamond and the Can-ada Dry can placed on the right point of the diamond,in effect removing the roll-down bias. A Schweppes gin-ger ale container was at the top of the diamond, and aSeven-Up container was placed at the bottom of thediamond. The third presentation reversed the positionsof the Canada Dry and Mountain Dew containers, withthe Canada Dry can to the left and the Mountain Dewbottle to the right, a counterbalancing procedure thatcontrolled for any left-to-right scanning bias. The Seven-Up container was at the top and the Schweppes gingerale container at the bottom of this slide. The fourth


presentation had the Canada Dry can on the top andthe Mountain Dew bottle on the bottom (Seven-Up onthe right, Schweppes on the left) and thus was also sus-ceptible to roii-down effects. The target experimentalbrands did not appear in any ofthe other presentations.

Attention Monitoring Apparatus. After viewing theexperimental tape the subject was led into a secondroom containing an Applied Science Laboratory cor-neai-reflection eye-tracking system (Young and Sheena1975). The subject was seated in front of a stimuluspresentation monitor. The monitor screen was locatedapproximately 34 inches from the subject (the horizon-tal plane ofthe eyes varied by the height ofthe subject).The eye-monitoring camera was located directly belowthe computer screen and was also approximately 34inches from the subject's eyes. The distance and anglefrom the eye to the computer screen and camera weredetermined by a factory calibration ofthe eye-trackingsystem.

After being seated, the eye-tracking system was ex-plained to the subject. Then, the subject was asked toplace his/her chin on a chin rest. At this point an eyecalibration was performed by having the subject lookat a video display of nine numbers arranged in a 3 X 3grid and then recording the characteristics ofthe eye asit looked at each position.' After the calibration wascompleted, the subject took a short break and then re-ceived instructions about viewing the soft drink dis-plays.

Measuring Conditioned Attention, After returningto the eye-tracking station, the subject was told thats/he would see a series of 24 slides, every second slidebeing a diamond arrangement of soft drink containers.Subjects were informed that tiller slides that had an "X"in the middle of the slide would appear prior to eachsoft drink slide. The subject was asked to look at the"X" until a stimulus slide appeared, then let his/hereye look at the soda containers in any fashion s/hepleased. Once the subject understood all ofthe instruc-tions, the presentation was started. Slides were thenpresented at five-second intervals. In a typical proce-dure, the time between the end ofthe commercial pre-sentation and the beginning ofthe test stimulus pre-sentation was approximately 10 minutes.

AnalysisData Preparation, The eye-tracking system re-

corded the distance between the center ofthe pupi! andthe corneal reflection every sixtieth of a second. Thisdistance was converted into an x, y coordinate that cor-responded to a location in the viewing area. These rawdata were then parsed into fixations by identifying se-quences of observations in which there was limited dis-

A description of the eye-tracking system and calibration meth-odology can be found in Young and Sheena (1975).

TABLE 1

AVERAGING VIEWING ORDER OF SOFT DRINK CONTAINERS:EXPERIMENT 1

First presentation:Mountain Dew forwardCanada Dry forward

Second presentation:'Mountain Dew forwardCanada Dry forward

Third presentation:"Mountain Dew forwardCanada Dry forward

Fourth presentation:Mountain Dew forwardCanada Dry forward

MountainDew

1.461.79

2.132.75

1.922.50

2.873.00

CanadaDry

3.213.00

2.912.33

2,752,45

2.332.13

Difference(SD)

-1.75(1,45)-1.21 (1.61)

-.79(1.74).42 (2.10)

-.83(1,71).05(1.97)

.54 (2.06),87(1,92)

NOTE.—There were 24 subjects per condition.'Repeated-measures analysis used data from the second and third presen-

tation.

placement of focus. Fixations were identified with analgorithm that identified sequences of data points thatrepresented less than a degree of movement within a117-millisecond time period. This manufacturer-de-fined algorithm for identifying fixations is a well-ac-cepted standard in eye-fixation research (Stark and Ellis1981).

A subject's fixation data were then combined withfiles specifying the location of each object in each dis-play. Thus, a summary of each subject's pattern of at-tention to each slide was constructed. The resultingsummary files contained information on the sequenceof fixations, the timing of fixations, and the number offixations on each soda container in the display.

Data Analysis and Results. Each brand of soda hada sequence score (1,2, 3, or 4) associated with the orderin which it was viewed. Nonviewed brands were as-signed a score of 4. We a priori limited our interest tothe second and third presentation but provide all datafor completeness.

Table 1 reports means that were used in the analysis.The analysis looked at the average order of viewing theMountain Dew container relative to the Canada Drycontainer (see, e.g.. difference scores in column 3 ofTable 1). A repeated-measures analysis of viewing or-ders in the second and third presentation showed a sig-nificant influence of the conditioning procedure ma-nipulation on the order of attention to Mountain Dewand Canada Dry. Across the two presentations. Moun-tain Dew was viewed 0.81 brands sooner ([-0.79+ -O.83J/2) than Canada Dry by the Mountain Dewforward conditioning group but 0.24 ([0.42 + O.O5J/2)brands later than Canada Dry by the C"anada Dry for-ward conditioning group, a significant difference


{F{2A5) - 3.33, p < .05, OJ' = .05).' On average. Moun-tain Dew was seen sooner (X = 2,13, X = 1.92) thanCanada Dry (X = 2.75, X = 2.50) by Mountain Dewforward conditioning group subjects {f(2,45) ^ 3.19, p< .05, aj= = .05) but later {X = 2.9\, X = 2.75) thanCanada Dry {X - 2.33, X - 2.45) by Canada Dry for-ward conditioning group subjects { '(2,45) = 1.60, p= .10,0)2 = .02)."

These results suggest that the conditioning procedurewas effective at getting the subject to look at the brandpresented with a forward conditioning proceduresooner. Although the effect sizes may seem small, vary-ing the order of attention to brands could have a sig-nificant influence on market share, when one considersthe impact of consideration set size on brand choice{Nedungadi 1990; Urban, Johnson, and Hauser 1985).

Contingency and Demand AwarenessNo attempt was made to assess contingency or de-

mand awareness during the study.^ Instead, indepen-dent assessments of contingency and demand awarenesswere made with unique groups of subjects drawn froma subject pool equivalent to that used for experiment1. For awareness to have influenced the order of atten-tion to the brands, the subject had to be able to (1)identify the treatment brand during training (a possibleconsequence of contingency awareness) and {2) use thisknowledge to direct attention to the treatment brandduring testing.

Our first test investigated whether subjects were ableto acquire information during the training session thatwould allow them to identify the treatment brand. Twogroups of subjects were tested. Twenty-six {27) subjectswere exposed to the sequence of 12 commercials con-taining the Mountain Dew (Canada Dry) forward con-ditioning commercial. Subjects were run in groups ofthree or fewer to encourage attention and to allow forrandom assignment. Immediately after viewing thecommercials, subjects were asked to complete the con-tingency-awareness Questionnaire.

The first question on the instrument asked subjectswhich commercial they liked more {see Appendix,question 1) in an effort to indirectly assess whether sub-jects were aware that one of the brands was being paired

^The analysis with the Initial viewing time means provided parallelresults.

•"The analysis could have been performed with the times at whicheach soft drink container was first viewed. The results of that analysisare similar to those presented. The only difference is that the CanadaDry, instead of the Mountain Dew. means are significantly differentin Ihe ninth presentation. We used the order of viewing data becausethey were less sensitive to subject idiosyncracies (i.e.. the length ofindividual fixations varies widely by subjects).

*The eye-tracking data collection procedure involved collectingdata for six projects at once. The data reported in this article werethe first data to be collected. A contingency- or demand-awarenessinstrument would have biased data collection for the other five pro-jects.

with a more positive US. The mean difference in likingdid not differ by experimental condition (F = 1.27, p> .10; see Table 2). Subjects were then asked to selectthe ad that did a better job of holding their attention,a question that indirectly assessed whether they couldidentify the brand that was being paired with a moreinteresting US {i.e., a form of contingency awarenessthat might later influence attention). Again, the meandifference did not vary by condition {/•' = .09, p> .10).

Next, a more focused measure of contingency aware-ness was administered. Subjects were asked to considerthe forward conditioning commercial. Subjects werereminded of the type of scenes that had been in thecommercial and then asked to recall any pattern in thesequence of the scenes, an open-ended measure of con-tingency awareness (question 3). Two questions later,subjects were also asked to express any contingencyawareness about the random scene sequence commer-cial, so as to have a comparative control {question 5).For each question, subjects were explicitly promptedto discuss whether some types of scenes always precededor followed other types of scenes. As shown in Table 2,the stated contingency awareness expressed for the for-ward conditioning commercial was comparable to thepseudo-contingency-awareness for the random condi-tioning commercial {in parentheses).

Finally, subjects were told that there was a 50-50chance that there was a pattern to the scenes they sawin the commercial. They were asked to indicate whetherthe pairing was (I) fun, active scenes predicted con-sumption scenes (US -* filler), {2) fun, active scenespredicted pictures of the product package {US -»• CS),{3) pictures of the product package predicted con-sumption scenes (CS -* filler), (4) pictures of the prod-uct package predicted fun, active scenes {CS -* US),(5) consumption scenes predicted fun, active scenes(filler -^ US), (6) consumption scenes predicted productpackage scenes {filler -* CS), (7) random order, or (8)don't know. Subjects were asked this question with ap-propriate labels (e.g., "Surfing scenes always came be-fore soda drinking scenes") and were asked to select asmany answers as applied. Again, this question was askedabout the forward (question 4) and the random {ques-tion 6) conditioning commercials. Responses are shownin Table 2 with the responses for the random condi-tioning commercial in parentheses.

The pattern of responses suggests that subjects werenot aware of the contingency between the CS and theUS. Levels of contingency awareness were 8 percent forthe Mountain Dew forward commercial subjects {IIpercent for the Canada Dry random commercial sub-jects) and 19 percent for the Canada Dry forward com-mercial subjects (15 percent for the Mountain Dewrandom commercial subjects; all Z < 1). Thus, reportsof contingency awareness seem to be at a chance level.Perhaps the most interesting aspect of the data was thelarge number of subjects that reported that interestingscenes (US) were followed by consumption {filler). Re-


TABLE 2

RESULTS OF CONTINGENCY- AND DEMAND-AWARENESSTESTS: EXPERIMENT 1

Contingency-awarenesstest:

Liking;Mountain Dew (1)-Canada Dry (9)'

Interest:Mountain Dew (1)-Canada Dry (9)'

Stated contingency:Mountain DewCanada Dry

Recognized contingency:US — fillerU S - C SCS— USCS -* fillerFiller — USFiller — CSRandom orderDon't know

Demand-awareness test:Was there a brand we

wanted you to lookat?

NoYes

Mountain DewCanada DryOther

Did you make a consciouseffort to look atone of the brandsfirst?

NoYes


Choose the brand wewanted you to lookat first:


Forced-choicecontingency ID:

Mountain DewCanada Dry

Mountain Dewforward

2.74(2.26)"

2.74(1.76)"

Mountain Dewforward^

Mountain CanadaDew Dry

11 (4)

55 (59)11 (19)8 (11)8 (0)

15 (11)4 (4)

15 (8)15 (19)

Mountain Dewforward

554535

010

7030150

15

550

45

5545

Canada Dryforward

2.12(1,73)"

2.57(2.10)'=

Canada Dryforward*^

Mountain CanadaDew Dry

(8) 18

(50) 46(15) 19(19) 15

(4) 19(4) 35(8) 8

(27) 4(15) 8

Canada Dryforward

42S8475S

64

as55

5811at

se42

"Item was a nine-point bipolar scale assessing ttie liking of or interest in theMountain Dew commercial relative to ttie Canada Dry commercial. Scale endpoints were labeled Mountain Dew (1) and Canada Dry {9).

''SD.''Pseudo-contigency-awareness values for the random conditioning commercial

appear in parentheses.

ports of this contingency were equivalent among theexperimental and random control groups, which sug-gests that subjects expected this format in soda com-mercials (i.e.. "You engage in an activity and then youhave a soda").

Even if subjects could not reliably differentiate be-tween the treatmetit and test tapes directly after thetraining session, differences in the commercials mayhave become more apparent in the course of the eye-tracking procedure. Our second test investigated thepossibility that demand awareness influenced the orderin which consumers attended to the brands. Subjects(n ^ 39) were put through the procedure used in ex-periment 1 (19 viewing the Canada Dry forward tape,,20 viewing the Mountain Dew forward tape) but wereinterrupted after viewing the presentation ofthe CanadaDry and Mountain Dew containers that had providedthe strongest evidence for conditioning effects (the sec-ond presentation). Immediately after viewing this slide,the subject filled out an instrument with the followingquestions: (1) "As you were looking at the last slide,did you think there was a brand that we wanted you tolook at?"; (la) "If so. which brand?'"; (2) "When youwere looking at the last sHde, did you make a consciouseffort to look at one ofthe brands first?"; (2a) "If so,which brand?'"; (3) "If you had to guess which brandwe wanted you to look at first, which brand would youchoose?"; (3a) "How confident arc you about thischoice?"; (4) "Either the Mountain Dew or the CanadaDry commercial was organized so that interesting scenesalways followed a picture ofthe product package. Whichcommercial was organized in this way?"; and (4a) "Howconfident are you about this choice?" All questions wereanswered while the critical test slide was present.

If some subjects were demand aware, we would expectthem to be able to select Mountain Dew (Canada Dry)as the brand we wanted them to look at or look at firstin the Mountain Dew (Canada Dry) forward condition(Table 2). Regardless of experimental condition, anequal proportion of subjects selected Mountain Dew asthe brand they thought we wanted them to look at andthe brand we wanted them to look at first (both Z< 1.0). Even when subjects were forced to guess theidentity ofthe brand we wanted them to view first, therewere no differences between conditions (both chi-squarenonsignificant at a = .10).^

Discussion

The first experiment provides evidence that attentioncan he influenced by a conditioning procedure. Subjects

*We performed a demand-awareness analysis. No subjects identifiedthe conditioned brand as the brand we wanted them to look at (ques-tion 1 a) or stated that they tried to look at this bratid first (2a), Twosubjects identilied the conditioned brand (la) and guessed with con-fidence thai this was the brand we wanted them to look at fit^t (3 atid3a).


receiving forward conditioning presentations of theMountain Dew commercial looked at the MountainDew container earlier than subjects presented with therandom control version of the same commercial. Sub-jects receiving forward conditioning presentations ofthe Canada Dry commercial looked at the Canada Drycontainer sooner than the subjects that viewed the ran-dom control version of the commercial. Differentialpreference for the two versions of each ad (forward andrandom) did not seem to account for the results, nordid contingency awareness of the CS-US pairing. Sub-jects seemed to be aware that US and filler scenes werebeing paired but had little awareness of the CS-US con-tingency and were just as likely to identify the randomcontrol commercial as having contingent CS and USscenes. Yet the first study did not investigate the mech-anisms responsible for the influence of the conditioningprocedure on attention, nor did it include measures ofassociative learning. We will address the issue of theunderlying mechanism first, leaving measurement ofassociative learning to experiment 3.

There are two competing viewpoints that may offera plausible explanation of how conditioning proceduresmight have influenced attention to a CS. The view ad-vanced thus far is that a basic learning system, a systemwe call the perceptual learning system, identifies poten-tial covariations within the environment and encour-ages a higher-order system to test these hypothesizedassociations by attending to these stimuli. This asso-ciation hypothesis has a parallel in the animal learningliterature. The signaling hypothesis assumes that thepairing of the CS and US allows an animal to learnabout the predictive value of the CS, especially whenthe US has positive or negative consequences (Hall andChannel! 1985; Hearst and Jenkins 1974). When theCS is presented in a novel context, subjects attend tothe CS to determine whether the previously learnedcontingency will hold (i.e., to assess the strength of therelationship; Pearce and Hall 1992). In effect, animalsactively gather information to test a CS-US associationhypothesis. Similarly, we expect that subjects in exper-iment I developed a perceptual "hypothesis" about aCS-US association that encouraged attention to the CSin the subsequent product display.

An alternative explanation for the results observedin experiment 1 relies on traditional Pavlovian condi-tioning theory, as opposed to associative learning the-ory, to explain the effects of the CS-US pairing. Theresults of experiment 1 could be attributed to the directresponse transfer of the attention (UR) associated withthe entertaining scenes {US) to the product (CS) thatpreceded those scenes. The transfer of the orienting re-sponse from the US to the CS could have occurred inthe same way a reflex response is transferred—as a pureconsequence of the pairing.

Developing a test to differentiate between these twoviews is difficult. For example, direct support for theassociative learning explanation requires that one

monitor the orienting response in two situations (Halland Channell 1985). First, as associative learning pro-gresses, attention to the CS during conditioning trialsshould decline because the information becomes re-dundant—the hypothesis has been generated and thepairing in that context provides no further information.Second, the subject should once again attend to the CSduring test trials, because an assessment of the veracityof the association is warranted. Although experiment Iprovided evidence for this second prediction, evidenceof declining attention during training would be difficultto generate in a laboratory setting. The heightened at-tention associated with the laboratory environmentwould make it difficult to observe a lack of attention tothe CS (i.e., it is the only stimulus on the screen).

One possible strategy for differentiating between theassociative learning explanation and the responsetransfer explanation of the differential attention to theCS in experiment 1 is to investigate ancillary predictionsassociated with each viewpoint. For example, the as-sociative learning explanation suggests that condition-ing should be strongest when alternative forms of theCS are used for each trial during training. Alternativeforms of the CS should encourage the perceptual systemto orient to the CS at each presentation of the CS, thusincreasing the likelihood that the perceptual system willbegin to learn the association between the CS and theUS {Pearce and Hall 1992). Maintenance of the ori-enting response should speed associative learning be-cause attention is instrumental in learning the CS-UScontingency {Ohman 1979; Vinogradova 1965). In op-erational terms, a commercial employing unique ver-sions of a product scene {CS) in each of the six trialsshould result in quicker associative learning than acommercial employing the same product scene.

In contrast, the response-transfer view predicts thatconditioning will be quicker when the CS is identicalduring all training trials {McSweeney and Bierley 1984;Pavlov 1927). An identical product scene {CS) will in-crease the likelihood that the orienting response to theUS, the more interesting of the two stimuli, will directlytransfer to the CS. This prediction can be traced to Pav-lov's definition of the CS as the sense organ stimulationresulting from contact with theCS{Kimble 1961; Pav-lov 1927). In effect, a constant CS allows the subject tolink the orienting response associated with the US to aconsistent sensory representation, thus speeding the rateat which conditioning occurs. In operational terms,conditioning should be quicker when an identicalproduct scene is used for each of the six trials. Thesealternative predictions were investigated in experi-ment 2.

EXPERIMENT 2The procedures used in experiment 2 were similar to

those used in experiment I. The primary differencesconsisted of a change in the experimental design and a


modification oflhe experimental materials. The exper-iment used a 2 {consistency of CS—varied vs. constant)X 2 (brand) design with an independent control group.The two tapes used in experiment 1 served as the variedCS tapes. Both commercials featured five unique CSscenes in the six CS positions. One CS scene was usedtwice in each commercial because the original com-mercials contained only five presentations of the brand.

The two constant CS tapes were constructed by re-placing the unique CS scenes in the varied CS versionsof the test commercials with an identical CS scene. Todo this, the longest CS scene from each varied CS com-mercial was copied and then spliced over the five re-maining CS scenes. In this way, all CS scenes in newlycreated constant CS versions of the test commercialsbegan and proceeded in an identical manner. Differ-ential lengths of the CS slots meant that there was somevariability in the length of each CS presentation, butthese lengths were identical to those found in the variedCS commercials. Fortuitously, the constant CS sceneswere also the cleanest pictures of the products. Theyfilled the entire screen and looked quite similar to thetest trial stimuli. The only potential problem was thatthe Canada Dry CS scene had ice cubes falling in frontof the container display (all Canada Dry CS scenes hadthis characteristic). This movement may have height-ened attention, but it also decreased the similarity be-tween the training and test CS. Again, this was an ac-cepted artifact resulting from the use of existingcommercials as stimuli.

A fifth condition consisted of a control tape contain-ing the random version of the Mountain Dew and Can-ada Dry commercials used in experiment 1. This ran-dom control would be used to provide a baseline toassess the direction of movement associated with theconditioning procedure. In experiment I. the sequenceof commercials shown to treatment group 2 includeda random version of the forward conditioning com-mercial shown to treatment group 1, and vice versa.This combining of treatment and control conditionsresulted in an overstating of the strength of the condi-tioning manipulation in experiment 1. A control groupcontaining no treatment commercials would be a moreconservative comparison group for assessing thestrength of the conditioning procedure because it wouldprovide attention measures that were unbiased by con-ditioning associated with other brands. This fifth con-dition was such a random control.

MethodSubjects. Subjects were 58 undergraduate juniors

and seniors enrolled in the senior author's ConsumerBehavior course and 120 undergraduate sophomores,juniors, and seniors from a Principles of Marketingsubject pool {64 women. 104 men). Consumer Behaviorcourse subjects participated in the experiment eightweeks prior to the class discussion of the conditioning

TABLE 3

AVERAGE VIEWING ORDER OF SODA CONTAINERS:EXPERIMENT 2

First presentation:Random controlVaried CS:

Mountain Dew forwardCanada Dry forward

Constant CS:Mountain Dew forwardCanada Dry forward

Second presentation:'Random controlVaried CS:

Mountain Dew forwardCanada Dt7 forward


Third presentation:'Random contrcrfVaried CS:

Mountain Dew forwardCanada Dry fonârd


Fourth presentation:Random controlVaried CS:

Mountain Dew for%wardCanada Dry forward

Constant CS:Mountain Dew fonwardCanada Dry forward

n

33

3333

3130

32

3333

3130

32

3332

3130

31

3232

3030

MountainDew

1.54

1,881.33

1 451,36

2.22

1.612.42

1.942.37

2.50

1.912.84

2.292.63

3,19

2,722.75

3.002.87

CanadaDry

3.36

3.153.33

3.323.17

2,84

3.182.51

3.002,70

2,43

2.762.10

2,842.37

2.03

2.472.19

2 102.30

Difference(SD)

-1.81 (1.53)

-1,27(1.64)-2.00(1.54)

-1.87(1,28)-1.80(1.35)

,62(1,58)

-1.58 (.93)-.09(2.00)

-1.06(1.82)-.33(171)

.06(1.98)

-.85(1,79).75(1,48)

-.55(2.01).27(1.76)

1.18(1.81)

,25(2.31).56(1.86)

90 (2.00).57(2.21)

"Repeated-measures analysis used data from the second and ttiird preseivtation.

literature. Conditioning material was not included inthe Principles of Marketing course.

Proeedures. All experimental procedures and anal-ysis techniques were identical to those used in experi-ment i. Missing subjects can be attributed to failuresin the presentation software {one subject), data lost be-cause of machine or operator failure during eye-tracking(seven subjects), and data lost because of subject move-ment during recording (variable). Every effort was madeto salvage data, hence the number of observations variesfor individual test presentations.

Analysis and Results

Data preparation was identical to that in experimentI. Each brand of soda was assigned a sequence score(1, 2. 3. or 4) associated with the order in which it wasviewed. As in experiment 1. we were interested pri-marily in attention to the brands presented on the sec-ond and third presentations, although the means of alldata are presented for completeness. Table 3 reportsthe average rank order of attention to brands by con-dition and the ditference between the rank order of at-


tention to the Mountain Dew container and the CanadaDry container.

The use of a random control design allowed for moreconservative tests of conditioning than were performedin experiment I. Inexperiment I. two treatment groupswere compared, with the conditioning procedure ma-nipulation having an independent influence in eachgroup. In experiment 2, a unique random control groupremoved any bias that might have occurred by pre-senting treatment and control commercials to the samesubjects. A priori planned contrasts involved comparingeaeh ofthe treatment groups' difference scores to therandom control group's difference score.

A repeated-measures analysis of viewing orders inthe second and third presentation showed a significantinfluence ofthe varied CS commercials on the order ofattention to Mountain Dew and Canada Dry. Acrossthe two presentations., Mountain Dew was viewed 0.93brands sooner ([0.94 + 0.91]/2) than Canada Dry bythe Mountain Dew varied CS conditioning group, ascompared with the random control group iE{2, 153)= 5.32, /)< .01. oj = .03). Canada Dry was viewed 0.61brands sooner ([-0.53 + -0.69]/2) than Mountain Dewby the Canada Dry varied CS conditioning group, ascompared with the random control group (F(2, 153)= 2.18,/7 = .05, oj = .01). On average. Mountain Dewwas seen sooner by the Mountain De^ varied CS con-ditioning group subjects {X = 1.61,^= 1.9 [) than bythe random control subjects {X = 2.22, X = 2.50;Fi2A5) = 5.17./? < .05, co- = .05). On average, CanadaDry was seen sooner by the Canada Dry varied CS con-ditioning group subjects {X = 2.51, X ^ 2.10) than bythe random control subjects (X = 2.84, X = 2.43;F{2,45)= 1.83,/)<.IO, oj = .01).

A repeated-measures analysis of viewing orders inthe second and third presentation showed an insignif-icant influence ofthe constant CS commercials on theorder of attention to Mountain Dew and Canada Dry.Across the two presentations. Mountain Dew wasviewed 0.52 brands sooner ([0.44 + 0.61 ]/2) than Can-ada Dry by the Mountain Dew constant CS condition-ing group, as compared with the random control group(F(2, 153) - 1.62,/7> .10, oj' < .01). Canada Dry wasviewed 0.25 brands sooner ([-0.29 + -0.2IJ/2) thanMountain Dew by the Canada Dry constant CS con-ditioning group, as compared with the random controlgroup (F(2, 153) = 0.31,p> .05, oj <.O1).

Discussion

The results of experiment 2 suggest that the associa-tive learning hypothesis may provide a better expla-nation ofthe influence ofthe conditioning procedureon attention to conditioned stimuli. The use of variedCS scenes during training supposedly increased atten-tion to the conditioned stimuli, which increased thelikelihood a CS-US association would develop. As aconsequence, subjects were more likely to attend to the

conditioned brand during the presentation ofthe prod-uct displays. The associative learning view would attri-bute attention to the CS to the subject's attempt to moreconfidently learn the CS-US contingency—a conse-quence of gathering more information on this associ-ation hypothesis in a novel context (Pearce and Hall1992). In contrast, the response transfer view wouldhave predicted that the constant CS would have allowedfor quicker conditioning. Although the constant CSmeans were in the proper direction, these differencesdid not reach significance.

Admittedly, the results of the second experimentcannot be considered overwhelming evidence for theassociative learning explanation ofthe role of attentionin conditioning. First, the influence ofthe varied CSconditioning procedure on attention was small, with cc-approaching only .03 in the most conservative tests.Second, a significant influence ofthe conditioning pro-cedure on attention was limited to Mountain Dew, oneof two of the brands studied, although the tests for Can-ada Dry did approach significance. Third, the manip-ulation was not a direct test ofthe associative learningand response transfer hypotheses but an indirect test ofancillary predictions of each explanation of the rela-tionship between conditioning procedures and atten-tion.

Despite these limitations, one must keep in mind thatexperiment 2 was an empirical test of two competingviewpoints in a domain in which stimulus materials,procedures, and contexts all influence the strength ofthe effects. Admittedly, the laboratory environmentprovided controls that would not have been availablein the natural environment, but there were also con-straints that limited the size ofthe conditioning effects.First, the treatment commercials employed weak un-conditioned stimuli, as is usually the case with ad ma-terial, and moderately familiar brands, a potential latentinhibition problem. Second, the experimental manip-ulation consisted of only 18 trials embedded in threecommercials, a small fraction ofthe number of repe-titions consumers encounter in a typical ad campaign.Third, the test materials provided for control in mea-surement but used only four brands. When one accountsfor the roil-down bias, it was often the case that thesubject was making a decision on whether to attend toone of two brands (i.e., the left or right side ofthe dia-mond), effectively placing a ceiling on the size of thetreatment effect. Thus, it is quite diflicult to determinewhether the effect sizes observed in experiment 2 arerepresentative ofthe relationship between conditioningprocedures and attention in the marketplace.

Although it is difficult to determine the robustnessofthe influence of conditioning procedures on atten-tion, it is possible to provide further support for theassociative learning hypothesis. The associative learningexplanation assumes attention is being allocated to theCS because some form of associative learning has takenplace, even though this learning is proposed to be oc-


curring in a basic, perceptual system. In contrast, theresponse transfer hypothesis predicts that the orientingresponse to the US will block other unconditioned re-sponses from transferring to the CS (Ohman 1983; Pav-lov 1927). Thus, finding evidence for a second responsethat is influenced by the conditioning procedure wouldprovide additional support for the associative learninghypothesis and evidence against the response transferhypothesis.

in attempting to find support for a CR. a broad viewof potential conditioned responses was considered.Traditionally, consumer researchers have concentratedon affective conditioning, an interest that can be tracedto Staats and Staats' original studies on associativelearning {Staats and Staats 1957. 1959). Yet, one mustremember that Staats and Staats viewed the affectivedimension as one of many meaning dimensions thatcould serve as a conditioned response (Osgood. Succi,andTannenbaum 1957: Staats 1968). Staats and otherswere able to show that the dimensions of affect, activity,potency, angularity, and roundness were all viable con-ditioned responses (e.g.. see Staats [1968] for review).Hence, a broad view of potential conditioned responsesneed not be limited to affective responses alone.

Given our uncertainty about the CR in experiments1 and 2 and our desire to view conditioning as a pro-cedure that could provide inputs into subsequent choiceprocesses, the focus ofthe third study was to investigatewhether repeated pairing of the brands and the USscenes would result in an association ofthe meaning ofthe US to the CS. We made no a priori judgments aboutthe specific meaning that would become associated withthe CS. Instead, we selected a procedure that allowedus to assess whether any meaning had been associated.

EXPERIMENT 3In attempting to provide evidence of a semantic CR,

it was important to recognize that there was uncertaintyassociated with the underlying meaning dimension thatwould be associated because ofthe CS-US pairing. Thesolution to this problem was to use a set of multidi-mensional scaling measures. A multidimensional scal-ing solution provides for a subject-driven assessmentof meaning association. Thus, a multidimensional scal-ing instrument would provide for an unconstrained as-sessment ofthe ability of conditioning procedures toassociate semantic responses. There was no attempt topredict the direction ofthe movement on the map. al-though it was understood that any movement shouldhave face validity (e.g.. the pairing of a brand with scenesof young people should encourage them to view thebrand as younger).

Stimuli and DesignThe stimuli were the Mountain Dew and Canada Dry

commercials used in experiment 1. Condition 1 con-

sisted of two repetitions of the forward conditioningversion ofthe Canada Dry commercial and two repe-titions ofthe random version ofthe Mountain Dewcommercial. Condition 2 consisted of two repetitionsof the forward conditioning version of the MountainDew commercial and two repetitions of the randomversion of the Canada Dry commercial. The controlcondition consisted of two repetitions of the randomversion of the Mountain Dew and the Canada Drycommercials.' As in experiment 1. the target commer-cials were embedded in a series of commercials for eightdifferent soft drinks, two experimental and six filler.

ProcedureStimulus Presentation, Subjects were 52 students

in the first author's class and were recruited prior totheir exposure to classical conditioning subject matter.The experiment was run during the second half of aregularly scheduled lecture. Subjects were randomly as-signed to experimental conditions and then led into oneof three rooms containing video screens. Subjectsviewed a tape appropriate to their experimental con-dition then returned to the classroom.

Dependent Measure. After returning to the class-room, the subjects were handed a dependent-measurequestionnaire containing 28 similarity scales. Subjectsused a nine-point scale anchored with "rarely" and "al-ways" to decide the likelihood that pairs of soft drinkswere consumed by the same person. This similaritymeasure was one of many that could have been usedand was selected because it was a good measure for sub-stitutability. a reasonable goal for a moderately knownsoft drink. Subjects saw all possible pairs of the eightsoft drinks. The order ofthe pairs was randomly deter-mined but constant for all subjects.

Analysis and ResultsEach subject's similarity data were submitted to a

metric multidimensional scaling anaiysis to generateindividualized sets of two-dimension coordinates forthe eight brands. Each subject's multidimensional scal-ing solution was then submitted to Procrustes IndividualDifference Scaling (PINDIS), a technique that allowsone to combine individual perceptual maps and tocompare combined maps. This technique (Borg andLingoes 1987) is similar to the more commonly usedIndividual Differences Scaling (INDSCAL) in that itallows for the derivation of a uniquely rotated, unit-scaled common perceptual space from which an indi-vidual subject's perceptual spaces can be derived via asimple transformation. The common unit scale places

'This experiment was run prior to experiment I and 2. The initialdesign also included two backward conditioning (US -- CS) treatmentgroups. These conditions are not pertinent to the key empirical issuein experiment 3 and were not included in the analysis.


all maps, individual and combined, in the same metric,so that commonly used ANOVA statistics can be usedto test the influence ofthe experimental manipulations.The brand location coordinates from individual per-ceptual spaces serve as input into the statistical tests.

Figure 2 presents the common perceptual map de-rived from PINDIS. The location for each brand in eachofthe three conditions is represented on the map. Thepositions ofthe brands in the perceptual map derivedfrom the control condition are shown in regular type-face. Brand positions for the perceptual map derivedfrom the Canada Dry forward condition are underlined.Brand positions for the perceptual map derived fromthe Mountain Dew forward condition are in bold. Alegend assigning brand names to numbers appears belowthe perceptual map.

Perceptions ofthe positions of brands 1-6, the non-conditioned control brands, are almost identical amongthe three conditions. Univariate ANOVA tests con-firmed that there were no differences in the represen-tation of these brands in perceptual space (all F < I).In other words, the positions of Crush. Minute Maid.Seven-Up, Slice, Schweppes. and Sunkist did not varybetween treatment and control groups. Canada Dry andMountain Dew were influenced by the experimentalmanipulations, primarily along the horizontal axis.Two-tail, univariate grand mean tests confirmed thatthere was a statistically significant difference in the rep-resentation of Canada Dry {F(2,49) = 3.49, p = .04,mean square error [mse] - 0.028) and Mountain Dew{m,49) - 4.46. p = .02, mse - 0.035). Thus, the po-sition of Mountain Dew and Canada Dry did vary be-tween treatment and control conditions.

Between-subject tests ofthe influence ofthe condi-tioning procedure were conducted to determine whetherthe distance between Canada Dry and Mountain Dewvaried by experimental condition. Scores representingthe distances between Canada Dry and Mountain Dewalong the horizontal axis were computed for the indi-viduals in each condition. The difference between Can-ada Dry and Mountain Dew was less than in the controlcondition, when subjects viewed the Canada Dry for-ward commercials (two-tail /• (1,49) = 4.43, p = .04).The difference between Canada Dry and Mountain Dewwas greater than in the control condition, when subjectsviewed the Mountain Dew forward commercials (two-tail F( 1,49) = 2.94, p - .09). Given the relative stabilityof the representations of the nonconditioned brands,these results can be viewed as evidence that the con-ditioning procedures influenced perceptions ofthe tar-get brands. A discussion of this influence is presentedbelow.

Discussion

The results suggest that the conditioning procedureinfluenced subjects' perceptions ofthe target consumersofthe soft drinks. As shown in Figure 2, forward con-

FIGURE 2

COMBINED PERCEPTUAL MAPS FROM EXPERIMENT 3

U.,iU

0.25 -

0,20 -

0.15 -

0.10 -

0.05 -

0.00 -

- 0,05 -

- 0 . 1 0 -

- 0 . 1 5 -

- 0.20 -

ctr

~55

" 5

i3

a

^^ QD.

•

MD UT)

4

2 22

1 1

' 6

MDL 6

4

'0,5 -0.3 -0.1 0.1 0.3

NOTE.Êxperimental conditions: regular typeface, control; underlined ty-peface, Canada Dry ftwward; boid typeface. Mountain Dew forward. Soft drinks:1. Crush; 2, Minute Maid; 3, Seven-Up; 4. Siice; 5, Schweppes; 6, Sunlîst; CD,Canada Dry; MD, Mountain Dew.

ditioning presentations were able to alter subjects' per-ceptions of the users of Mountain Dew and CanadaDry on a perceptual map. Admittedly, there was noprediction about the expected direction of the brandmovement on the map, since there could be no a prioriprediction ofthe dimensional space. Yet the map, andthe movement of the target brands on the map, doeshave a fair amount face validity. The X-axis could beinterpreted as maturity, with Schweppes Ginger Ale andCanada Dry representing the most mature soft drinks.Seven-Up a moderately mature soft drink, and Moun-tain Dew and the four fruit sodas the least mature softdrinks. Presenting Canada Dry scenes prior to picturesof young couples having fun encouraged consumers toperceive it to be more like Mountain Dew and the otherflavored soft drinks, as illustrated by the Canada Dryforward conditioning map (underlined typeface). Pre-senting Mountain Dew scenes prior to scenes of teenageactors having fun encouraged consumers to perceive itto be less like Canada Dry and more like the other fla-vored soft drinks, as illustrated by the Mountain Dewforward conditioning map (bold typeface).

The results of experiment 3 can be taken as evidencethat semantic meaning that accompanies perception ofthe US (e.g., couples in their twenties or teenagers) hasthe potential to associate to the CS. The visual materialpresented in each of the experimental conditions wasidentical, the only difference between the treatment andcontrol conditions being the order of the scene se-quencing. Hence, differences in the representation ofthe brands on the perceptual map can be directly at-


tributed to the conditioning procedure. And. eventhough these effects were subtle, it may be that the se-mantic conditioning occurred with little contingencyawareness. As the postexperimental inquiry resultsprovided after experiment 1 illustrate, the conditioningprocedures used in the television commercials were inno way transparent.

The successful conditioning of a semantic responsein experiment 3 provides additional support for the as-sociative learning explanation ofthe influence of con-ditioning procedures on attention to the CS during sub-sequent exposure. The associative learning explanationofthe differential attention to the conditioned stimuliin experiments I and 2 assumed that there was someform of associative learning during the presentation ofthe ads. The formation of an associative hypothesisduring the training session encouraged the subjects toattend to the CS during subsequent exposure, an ap-parent attempt to increase their confidence in the CS-US contingency. This associative hypothesis was alsoable to inffuence responses on a multidimensional scal-ing measure of meaning association. Thus, the condi-tioning procedure has multiple benefits, a finding con-sistent with the associative learning explanation butinconsistent with the direct response transfer explana-tion ofthe influence of a conditioning procedure.

GENERAL DISCUSSIONThe results of the three studies suggest that condi-

tioning procedures may be appropriate for generatingmultiple responses. Experiment 3 provides evidencethat conditioning procedures can be used to associatemeaning with brands, and experiments I and 2 illustratethat, to the extent consumers begin to learn an associ-ation between a CS and a US, their attention to the CSwill be enhanced during subsequent exposure. Together,the three experiments provide evidence for McSweeneyand Bierley's (1984) prediction that successful condi-tioning procedures encourage approach behavior, re-gardless ofthe CR, because they encourage a person togather additional information on the association thatis being learned. Thus, it seems possible to use the se-quence of scenes in an ad to encourage attention to aCS, then allow for the CR to serve as an input into achoice behavior. Furthermore, this strategy is likely tooperate when involvement is sufficient to encourage at-tention to the CS and US during ad exposure but in-sufficient to promote the conscious awareness of a CS/US contingency (see Allen and Janiszewski 1989).

If we view conditioning as a multiple benefit proce-dure that has the potential to influence attention, af-fective responses and product perceptions provide astrategy for using conditioning procedures to benefitlesser-known brands. For many moderately knownbrands, part-list cuing interference limits the likelihoodthe brand will enter the consumer's consideration setvia memory retrieval (Alba and Chattopadhyay 1986).

Even when the consumer confronts the display and at-tempts to update his/her awareness and considerationset, shelf allocation constraints discourage attention tothe moderately known brands. Thus, to have a realisticchance at being purchased for purposes of trial or varietyseeking, the moderately known brand must first capturethe consumer's attention. But capturing attention is notsufficient to motivate selection of a brand. The evidenceprovided above suggests that conditioning proceduresmay encourage a consumer to attend to a brand and,at the same time, make semantic information availablefor comparing choice alternatives. Hence, simple as-sociative learning techniques have the potential to in-fluence brand choice in a low-involvement purchasevia attention and semantic associations to the brandpackage.

Even though conditioning Is a multiple-benefits pro-cedure, one must keep in mind that it is a procedure,not a process. As a procedure, it creates a learning en-vironment, but it does not exclude one or another pro-cess from operating. Thus, there is no reason to assumethe processes and mechanisms supporting "condi-tioned" reflexive, attentive, affective, and semantic re-sponses are in any way equivalent. For example. Pav-lov's original work (1927) investigated the reflexresponses associated with the direct stimulation ofmuscles and glands. His procedures investigated the di-rect transfer of a physiological response. As such, con-tingency awareness was not a consideration. In contrast,consumer researchers have often investigated affectiveresponses associated with the activation of memory, anassociative learning procedure in which contingencyawareness is difficult to avoid (Allen and Janiszewski1989; Shimp et al. 1991b). Consumer researchers oftenuse a simple, two-stimulus learning environment, aprocedure that encourages the higher-order learningsystem to become aware ofthe CS-US contingency.

In this article, a conditioning procedure is used toinfluence an associative learning process that is hy-pothesized to rely on a basic perceptual learning system,a system that can selectively choose stimuli from acomplex environment, form hypotheses about their re-lationships, then act to gather the information that willallow higher-order learning systems to become awareof and confirm these relationships (Holyoak etal. 1989).This perceptual system can learn and respond to simpleassociations, covariations, and fluency cues but doesnot rely on conscious control mechanisms for thislearning {see Bruner 1992; Hall 1991; Holyoak et al.1989; Reber 1989). In other words, until the perceptualsystem has provided enough information for the higher-order system to recognize a contingency, awareness willremain limited but learning will proceed. Therefore,people can begin to learn a CS-US contingency andrespond in a manner that is consistent with this learningbut still not be contingency aware. It is this type oflearning that is most likely to occur during the low-involvement processing of ads.


Viewing conditioning as a procedure, not a process,that has the potential to affect a variety of learningmechanisms suggests that we should begin to questionthe generalizability of conditioning findings acrossstimuli, procedures, contexts, and subject populations(Shimp, Hyatt, and Snyder 1991). For example, con-sider the possibility that animals have a perceptual sys-tem that is far less successful at selecting and organizinginformation within the environment. One method ofcompensating for a system that has difficulty recogniz-ing CS/US associations is to organize the learning en-vironment so that these associations are obvious. Infact, animals do require constant stimuli, similar train-ing and test environments, previously nonpredictiveconditioned stimuli, and tight conditioning procedures.These procedural conventions replace the perceptualscreening and organization system that allows the or-ganism to identify the salient stimuli within the envi-ronment and to recognize the associations betweenthem.

In contrast to animals, humans have a perceptualsystem that is efficient at selecting and organizing in-formation within the environment and a higher-orderprocessing system that is quick to test and confirm hy-potheses (Bruner 1992). The efficiency ofthe perceptuallearning system suggests that procedural consistenciesthat motivate associative learning are likely to be of lessvalue than procedural variance that encourages gen-eralizability of the learned covariation. Thus, the useof a varied CS (experiment 2) may be quite beneficialfrom a generalizability perspective but not at all det-rimental from an associative learning perspective. Thehuman perceptual system is designed to handle vari-ability in stimuli and may learn more efficiently andeffectively when this variability is present (Pribram andMcGuinness 1975).

Our discussion ofthe apparent disparity between an-imal and human conditioning results is not meant toimply that the animal literature does not provide nu-merous insights into the potential role of conditioningprinciples in human learning. Instead, it is meant toserve as a reminder that there are functional learningdifferences in species populations and that these differ-ences need to be recognized prior to developing a con-ditioning program to influence consumers (Bitterman1965). These differences may relate to differences insensory, motor, or motivational factors, but they mayalso relate to differences in response to context changes,procedural constraints, and CS-US characteristics. Forexample, conventional wisdom is that establishedbrands are more difficult to condition because of latentinhibition (McSweeney and Bierley 1984). Yet, Shimpet al. (1991 b) were able to condition affective responsesto moderately familiar brands, and our results providesimilarevidenceconcerningsemantic responses. Giventhe context-dependent learning capabilities of humans,latent inhibition may not constrain conditioning to the

extent it does with animal subjects (see Swartzentruberand Bouton 1992).

It is because of the flexibility in human learningmechanisms that we find inconsistencies in the con-sumer conditioning literature. Humans are highly sen-sitive to the procedures used to communicate infor-mation and the measures of the impact of thesecommunications. Slight changes in procedure can oftenhave a significant influence on the learning that isachieved in a session, quite possibly a consequence ofdifferential participation by alternative learning systems(see e.g., Gorn 1982; Kellaris and Cox 1989; cL Shimpet al. 1991a). One example ofthe consequences of al-ternative learning systems is the apparent necessity ofcontingency awareness for successful conditioning inthe procedures used by Allen and Janiszewski (1989)and Shimp et al. (1991b) but the apparent irrelevanceof contingency awareness for successful conditioningin the procedures of Baeyens. Eelen, and Van den Bergh(1990) and the above studies. The conditioning litera-ture is an empiricist literature in which procedural ca-veats (e.g.. CS preexposure, US preexposure. URstrength, interstimulus interval, intertrial interval,temporal priority, number of trials, training environ-ment, test environment, and CS consistency) oftendominate practical goals (e.g., a measurable change ina significant behavior). As the research focus shifts froman emphasis on procedure to an emphasis on learningsystems, many ofthe apparent inconsistencies withinthe literature may be resolved.

If the conclusions drawn from behavioral studies aredependent on the procedures used to produce these ef-fects, then we have identified the limitation and benefitsofthe studies reported within this article, as well as ofthe consumer conditioning literature in general. Be-havioral research findings are by definition context- andprocedure-dependent, a bias that is likely to be mag-nified by the complexity and flexibility of human learn-ing mechanisms. Thus, our findings are dependent onthe selection of the conditioned stimuli (moderatelyknown), the selection of the US scenes (dynamic,arousing, entertaining), UR strength (relatively limited),interstimulus interval (none), intertrial interval (vari-able, approximately five seconds), temporal priority(CS -^ US), number of trials (12 or 18). training en-vironment (dynamic, involving, laboratory), test en-vironment (static, involving, laboratory), test distractors(other moderately popular brands), CS consistency(varied), and subject pool (college students, target mar-ket). All of these procedural parameters are constraintson the generalizability ofthe results. Yet, unlike manypast conditioning studies, we have tried to be explicitabout the learning mechanisms that are being influ-enced by the conditioning procedure and the boundariesfor which we expect these results to hold. Familiarbrands, actual commercials, an absence of intertrial in-tervals, and novel test environments are all virtuallyunavoidable in the natural marketplace; thus, it is ad-


vantageous that our studies incorporated these char-acteristics.

The procedural constraints on our findings, and thoseinvestigating consumer conditioning in general, wouldseem to imply that the generalizability of all behavioralfindings within consumer research is limited. This con-clusion is valid, but we offer three qualifications. First,as conditioning procedures become more ecologicallyvalid and procedures converge on those observed in themarketplace, confidence in the generalizability of theconditioning results will be enhanced {Lynch 1982).Second, as a diversity of procedures are used to provideadditional evidence for a conditioned response, as isthe case with affective conditioning, confidence in thegeneralizability of the effect should increase (Calder,Philips, and Tybout 1982). Of course, this assumes thatreplications represent the use of alternative methods,measures, and procedures—a necessary prerequisite ifclassical conditioning, and associative learning mech-anisms in general, is to be integrated into the cognitivelearning paradigm. Finally, as researchers begin to ex-plicate the learning systems or processes that are sen-sitive to the conditioning procedure, the conditioningliterature will be able to establish a link to cognitivelearning findings. Mechanisms and processes that sup-port covariation learning, associative learning, attitudeformation, memory formation, and the allocation ofattention are all influenced by conditioning procedures.Perhaps it is time that conditioning researchers beginto recognize these relationships and to isolate the ad-vantages of a conditioning procedure relative to a morecognitively based learning program.

APPENDIX

Contingency Awareness AssessmentInstrument

The series of commercials you just saw included twocommercials that were repeated three times each{Mountain Dew and Canada Dry). We would like toknow your impression of these advertisements.1. Which commercial did you like more? If you liked

the Mountain Dew commercial much more than theCanada Dry commercial, circle T . If you liked theCanada Dry Commercial much more than theMountain Dew commercial, circle '9'. If you likedeach commercial about the same, circle '5'. Othervalues represent opinions between these extremes.

Liked Mountain Dew Liked ihcm Liked Canada Drymore than Canada Dry equally well more than Mountain Dew

1 2 3 4 5 6 7 8 92. Which commercial was better at holding your in-

terest?

Mountain Dew Both did well Canada Dry1 2 3 4 5 6 7 8 9

3. We would now like you to think of the MountainDew ad. Recall that the commercial consisted of afast paced sequence of scenes (by 'scenes* we meansurfing scenes, pictures of teenagers drinking Moun-tain Dew, pictures of the product, etc.). Did you no-tice a pattern to the scenes? If so, describe it.

If you did notice a pattern to the scenes in theMountain Dew ad, was it the case that some type ofscene always followed or came before another typeof scene? If so, please describe?

4. Depending on your experimental condition, there isa 50-50 chance that there was a pattern to the scenesyou saw in the Mountain Dew commercial. Pleaseindicate your best guess about this pattern by circlingthe appropriate response (you may circle more thanone).a. Surfing scenes always came before soda drinking

scenes.b. Surfing scenes always came before pictures of the

product.c. Pictures of the product always came before surfing

scenes.d. Pictures of the product always came before soda

drinking scenes.e. Soda drinking scenes always came before surfing

scenes.f. Soda drinking scenes always came before pictures

of the product.g. I think the scenes I saw were random in their or-

der.h. I honestly don't know.

5. We would now like you to think of the Canada Dryad. Recall that the commercial consisted of a fastpaced sequence of scenes (by "scenes' we mean pic-tures of people having fun, pictures of people drink-ing Canada Dry. pictures of the product, etc.). Didyou notice a pattern to the scenes? If so, describe it.If you did notice a pattern to the scenes in the CanadaDry ad, was it the case that some type of scene alwaysfollowed or came before another type of scene? If so,please describe?

6. Depending on your experimental condition, there isa 50-50 chance that there was a pattern to the scenesyou saw in the Canada Dry commercial. Please in-dicate your best guess about this pattern by circlingthe appropriate response (you may circle more thanone).a. People-having-fun scenes always came before

soda drinking scenes.b. People-having-fun scenes always came before

pictures of the product.c. Pictures of the product always came before peo-

ple-having-fun scenes.d. Pictures of the product always came before soda

drinking scenes.e. Soda drinking scenes always came before people-

having-fun scenes.

188 JOURNAL OF OONSUMER RESEARCH

f. Soda drinking scenes always came before picturesof the product.

g. I think the scenes I saw were random in their or-der.

h. 1 honestly don't know.

[Received July 1992. Revised October 1992.]

REFERENCESAlba, Joseph W. and Amitava Chattopadhyay (1986), "Sa-

lience Effects in Brand Recall," Journal of MarketingResearch. 23 (November). 363-369.

Allen, Chris T. and Chris A. Janiszewski (1989), "Assessingthe Role of Contingency Awareness in Attitudinai Con-ditioning with Implications for Advertising Research,"Journal of Marketing Research. 26 (February), 30-43.

and Thomas J. Madden (1985), "A Closer Look atC]ass\ca\ Conditioning" Journal of Consumer Research,12 (December), 301-315.

Baeyens. Frank. Paul Eelen, and Omer van den Bergh (1990),"Contingency Awareness in Evaluative Conditioning: ACase for Unaware Affective-Evaluative Learning," Cog-nition and Emotion. 4(1), 3-18.

Bierley, Calvin, Frances K. McSweeney. and Renee Van-nieuwkerk (1985). •"Classical Conditioning of Preferencesfor Stimuli," Journal of Consumer Research, 12 (De-cember). 316-323.

Bitterman. M. E. (1965). "Phyletic Differences in Learning,"American Psychologist. 30 (June). 396-410.

Borg. Ingwer and James C. Lingoes (1987), MultidimensionalSimilarity Structure Analysis. New York: Springer.

Brown, Paul L. and Herbert M. Jenkins (1968), "Auto-shapingof the Pigeon's Key-Peck," Journal of the ExperimentalAnalysis of Behavior. 11 (January), 1-8.

Bruner. Jerome (1992), "Another Look at New Look 1,"Atnerican Psychologist. 47 (June), 780-783.

Calder. Bobby J., Lynn W. Philips, and Alice M. Tybout(1982), "The Concept of External Validity." Journal ofConsumer Research. 9 (December), 240-244.

Hall, Geoffrey (199!). Perceptual and Associative Learning.Oxford: Clarendon.

and Stephen Channell (1985), "Differential Effects ofContextual Change on Latent Inhibition and on the Ha-bituation of an Orienting Response," Journal of Exper-imental Psychology: Animal Behavior Processes. II(June), 470-48 L

Hearst, Eliot and Herbert M. Jenkins (1974). Sign-Tracking:The Stimultts-Reinforcer Relation and Directed Attention.Austin, TX: Psychonomic Society.

Holyoak, Keith J., Kyunghee Koh. and Richard E. Nisbett(1989), "A Theory of Conditioning: Inductive Learningwithin Rule-based Default Hierarchies," PsychologicalReview. 96 (April), 315-340.

Kaye, Helen and John M. Pearce (1984), "The Strength ofthe Orienting Response during Blocking," QuarterlyJournal of Experimental Psychology. 36B (May), 131-144.

Kellaris, James J. and Anthony D. Cox (1989), "The Effectsof Background Music in Advertising: A Reassessment,"Journal of Consumer Research. 16 (June), 113-118.

Kimble, Gregory A. (1961), Conditioning and Learning. NewYork: Appleton-Century-Crofts.

Lynch, John G., Jr. (1982), "On the External Validity of Ex-periments in Consumer Research," Journal of ConstimerResearch. 9 (December), 225-239.

Lynn, Richard (1966). Attention. Arousal and the OrientationReaction. Oxford: Pergamon.

McSweeney, Frances K. and Calvin Bierley (1984). "RecentDevelopments in Classical Conditioning," Journal ofConsumer Research. II (September), 619-631.

Nedungadi, Prakash (1990). "Recall and Consumer Consid-eration Sets: Influencing Choice without Altering BrandEvaluations." Jourtial of Consumer Research. 17 (De-cember), 263-276.

Ohman, Arne (1979), "The Orienting Response, Attention,and Learning: An Information Processing Perspective,"in The Orienting Reflex in Hutnans. ed. H. D. Kimbleet al., Hillsdale, NJ: Eribaum. 443-471.

(1983). "The Orienting Response during PavlovianConditioning," in Orienting and Hahititation: Perspec-tives in Human Research, ed. David Siddle, New York;Wiley, 315-369.

Osgood, Charles E., George J. Succi. and Percy H. Tannen-baum (1957), The Measurement of Meaning. Urbana:University of Illinois Press.

Pavlov, Ivan P. (1927), Conditioned Reflexes. Oxford: OxfordUniversity Press.

Pearce, John M. and Geoffrey Hall (1992). "Stimulus Signif-icance, Conditionability, and the Orienting Response inRats," in Attention and Information Processing in Infantsand Adults: Perspectives from Human and Anitnal Re-search, ed. Byron A. Campbell et al.. Hillsdale. Ni: Eri-baum, 137-162.

Pribram, Karl H. and Diane McGuinness (1975), "Arousal,Activation, and Effort in the Control of Altention," Psy-chological Review. 82 (March). 116-149.

Razran. Gregory (1961), "The Observable Unconscious andthe Inferable Conscious in Current Soviet Psychophysi-ology: Interoceptive Conditioning, Semantic Condition-ing, and the Orienting Reflex," Psychological Review. 68(March), 81-147.

Reber. Arthur S. (1989), "Implicit Learning and TacitKnowledge." Journal of Experimental Psychology: Gen-eral. 118 (September), 219-235.

Shimp, Terence A. (1991), "Neo-Pavlovian Conditioning andIts Implieations for Consumer Theory and Research,"in Handbook of Consumer Behavior, ed. Thomas S.Robertson and Harold H. Kassarjian, Englewood Cliffs,NJ: Prentice Hall, 162-187.

Eva M. Hyatt, and David J. Snyder (1991 a), "A Crit-ical Appraisal of Demand Artifacts in Consumer Re-search," Journal of Consumer Research. 18 (December),273-283.

Elnora W. Stuart, and Randall W. Engle (1991b). "AProgram of Classical Conditioning Experiments TestingVariations in the Conditioned Stimulus and Context,"Journal of Consumer Research. 18 (June), 1-12.

Sokolov, Evgenni N. (1963), Perception and the ConditionedReflex. Oxford: Pergamon.

Spinks, John A. and David Siddle (1983), "The FunctionalSignificance of the Orienting Response," in Orienting andHabituation: Perspectives in Human Research, ed. DavidSiddle, Chichester: Wiley. 237-313.

Staats. Arthur W. (1968), Learning. Language and Cognition.New York: Holt. Rinehart & Winston.


and Carolyn K. Staats (1959). "Effect of Number ofTrials on the Language Conditioning of Meaning,"Journal of General Psychology. 6! (October), 211-223.

Staats. Carolyn K. and Arthur W. Staats (1957). "MeaningEstablished by Classical Conditioning," Journal of Ex-perimental Psychology. 54 (July). 74-80.

Stark. Lawrence and Stephen R. Ellis (1981). "ScanpathsRevisited: Cognitive Models Direct Active Looking."in Eye Movements: Cognition and Visual Perception.ed. Dennis F. Fisher et al.. Hillsdale. NJ: Eribaum., 193-226.

Stuart. Elnora W,. Terence A. Shimp. and Randall W. Engle(1987), "Classical Conditioning ofConsumer Attitudes:Four Experiments in an Advertising Context." Jottrnalof Consumer Research. 14 (December). 334-349.

Swartzentruber. Dale and Mark E. Bouton (1992). "ContextSensitivity of Conditioned Suppression Following

Preexposure to the Conditioned Stimulus." AnimalLearning and Behavior. 20 (2). 97- 108.

Urban, Joel L., Philip L. Johnson, and John R. Hauser (1985),"Testing Competitive Market Structures." .MarketingScience. 3 (Spring). 83-112. •

Vinogradova. O. S. (1959). "The Role ofthe Orienting Reflexin the Formation of Conditioned Connections in Man,"in The Orienting Reflex and Prohlems of Higher NervousActivity, ed. Evgenni N. Sokolov. Moscow: AkademiiaPedagogicheskikh Nauk. 86-160.

(1965). "On the Dynamics ofthe Orienting Reflex inthe Course of Closure of a Conditioned Connection." inThe Orienting Rejlcx and Exploratory Behavior, ed. L.G. Voronin et al., Moscow: Akademiia PedagogicheskikhNauk RSFSR. 45-53.

Young. Laurence R. and David Sheena (1975). "Eye-Move-ment Measurement Techniques," American Psychologist,30 (March), 315-330.

Documents

The Influence of Classical Conditioning Procedures on Subsequent Attention … · 2017-05-05 · cessful classical conditioning, assisting the subject in learning an association between