THE BASIS OF SUPERSTITIOUS BEHAVIOR: CHANCE …bsl/Papers from lab/Timberla_1985.pdf · the basis...

1985, 44, 279-299 NUMBER 3 (NOVEMBER)JOURNAL OF THE EXPERIMENTAL ANALYSIS OF BEHAVIOR

THE BASIS OF SUPERSTITIOUS BEHAVIOR:CHANCE CONTINGENCY, STIMULUS SUBSTITUTION,

OR APPETITIVE BEHA VIOR?

WILLIAM TIMBERLAKE AND GARY A. LUCAS

INDIANA UNIVERSITY

This research examined three explanations for the "superstitious" behavior of pigeons under frequentfixed-time delivery of food: accidental response-reward contingency, stimulus substitution, and elicitedspecies-typical appetitive behavior. The behavior observed in these studies consisted of occasionalpostfood locomotion away from the food hopper, and a predominant pattern of activity directed towardthe hopper wall (wall-directed behavior), including approaching, stepping side to side, scratchingwith the feet, bumping with the breast, pendulum movements of the extended neck, and head bobbing,though not pecking. The consistency of these behavior patterns argued against explanation by acci-dental response contingencies, and the complexity of behavior was incompatible with the classicstimulus-substitution account. These studies also showed that: (1) response contingencies and priorstimulus experience can modify wall-directed behavior, but within definable limits; (2) pecking some-times can be obtained in birds of specific strains, and by providing extended training; (3) placing thehopper in the floor at the center of a large chamber replaces wall-directed behavior with circling ina manner that resembles ground foraging for food. We conclude that superstitious behavior underperiodic delivery of food probably develops from components of species-typical patterns of appetitivebehavior related to feeding. These patterns are elicited by a combination of frequent food presentationsand the supporting stimuli present in the environment.

Key words: superstitious behavior, appetitive behavior, behavior systems, stimulus substitution,fixed-time schedules, wall-directed behavior, pecking, pigeons

When pigeons receive food at frequent fixedtimes independent of their behavior, they rap-idly develop a set of relatively stereotyped re-sponses. These responses are typically de-scribed as "superstitious" because they emergein the absence of actual environmental contin-gencies, presumably like superstitious behav-ior in humans (Skinner, 1948). The point ofthe present studies was to explore the formand causation of such behavior in pigeons. Wewill use the term superstitious (without quotes)to refer to behavior that emerges under a re-sponse-independent schedule, but withoutpresumption as to its causation.

The two predominant explanations of su-perstitious behavior in pigeons are the acci-

This research was supported in part by PH8 Grants807 RR 7031, MH 39345 and by N8F Grants 82-10139and 84-11445. We thank Peter Harvey, Dean Hill, Rob-ert jeffers, and David Martino for their assistance andDon Gawley, Nancy Innis, and john 8taddon for com-ments. We also thank Nancy Innis and Peter Balsam forsharing with us videotapes and films of their subjects.Reprints may be obtained from William Timberlake, De-partment of Psychology, Indiana University, Blooming-ton, Indiana 47405.

dental response-contingency hypothesis ofSkinner (1948) and the stimulus-substitutionhypothesis as outlined by Staddon and Sim-melhag ( 1971) .According to Skinner, super-stitious behavior is an operant that occurs un-der the response-independent presentation ofreward because accidental (unprogrammed)juxtapositions of reward and response estab-lish a positive feedback loop. Initial accidentalresponse-reward contingencies raise the prob-ability of a response, thereby increasing thelikelihood of further accidental response-re-ward contingencies, etco The experimentalsituation Skinner used to demonstrate super-stitious behavior involved the response-inde-pendent delivery of grain to a hungry pigeononce each 15 So Skinner reported that pigeonsrapidly developed a variety of characteristicindividual patterns of behavior, summarizedin Table 1. Most of these patterns were stableover time, although Skinner noted that somepatterns drifted in topography, presumably asthe result of continued feedback from acciden-tal juxtapositions of food and behavior o

In contrast to Skinner's (1948) view, Stad-don and Simmelhag (1971) argued that su-perstitious behavior patterns are members of

279

WILLIAM TIMBERLAKE and GARY A. LUCAS280

Table 1

Category comparison: Skinner (1948) and Staddon andSimmelhag (1971).

Staddon & SimmelhagSkinner

TurnHead and neck pendulum

Step, hop (dance)

Thrust head in upper cornerToss headBru~h at floor with beakWalk about chamber

Quarter circleHead movements along

magazine wall

Moving along magazinewall

Beak to ceilingDizzy motionPecking floorLocomotionMagazine wall

Pecking keyPecking (wall)Pecking window wall

Flapping wingsWindow wall

PreeningHead in magazineHead to magazine

and others have reported more general casesin which animals continued to behave as if acontingency existed between their behavior andreward after the contingency had been re-moved. On the other hand, Reberg, Mann,and lnnis (1977), Reberg, lnnis, Mann, andEizenga (1978), Blaine, lnnis, and Staddon(1980), lnnis, Reberg, Mann, Jacobson, andTurton (1983), and lnnis, Simmelhag-Grant,and Staddon (1983) reported data that ex-tended the stimulus-substitution/ elicited-be-havior interpretation to longer interfood in-tervals, to water presentation, and to rats assubjects (although similar results were notfound by Fenner, 1980, who reported verylittle pecking by pigeons on fixed-time sched-ules}.

The main purpose of the present set of ex-periments was to investigate the obvious dis-parity between both the empirical results andthe theoretical interpretations of Skinner(1948) and Staddon and Simmelhag (1971).The first experiment established the existenceof a reliable form of superstitious behaviorunder a short fixed-time schedule of food pre-sentation. Experiments 2 to 5 tested more di-rectly the importance of response contingen-cies in determining superstitious behavior .Experiments 6 to 8 investigated the determi-nants of the terminal behavior of pecking re-ported by Staddon and Simmelhag. The re-sults of these experiments strongly suggestedthe alternative hypothesis that superstitiousbehavior is comprised of elicted componentsof species-typical appetitive behavior relatedto feeding. The final study and discussion be-gan an analysis of superstitious behavior fromsuch an approach (Timberlake, 1983a, 1983b,1983c, 1984; see also Staddon & Simmelhag,1971; Staddon, 1983).

GENERAL METHOD

SubjectsUnless otherwise noted, pigeons in all ex-

periments were naive female White Carneauxretired breeders (5 to 7 years old), obtainedfrom the Palmetto Pigeon Plant. The birdswere individually housed in 30.5 by 30.5 by30.5-cm sheet-metal cages with 2.54 by 2.54-cm hardware cloth fronts and stainless steelmesh floors. A 12:12-hr light: dark cycle wasin effect and water was freely available. Dur-ing experiments the pigeons were maintainedat 77.5% of their ad-lib body weights by sup-

a class of terminal responses elicited in antici-pation of food rather than accidentally rein-forced. In support of this view, they reportedthat pigeons exposed to short fixed-time (FT)or variable-time (VT) schedules developed acommon terminal response of pecking themagazine wall prior to the delivery .of food.They argued that the commonality of thepecking response favored an explanation basedon elicitation rather than on accidental re-sponse contingencies. They also noted that theoccurrence of pecking was readily explainedby the classic stimulus-substitution hypothesis(Pavlov, 1927), according to which condi-tioned responses are formed from componentsof the unconditioned reaction to the uncondi-tioned stimulus (in this case, the food).1 Priorto the terminal response of pecking, the pi-geons also engaged in various interim behav-ior patterns apparently unrelated to food. Thecategories of behavior recorded by Staddon andSimmelhag are also noted in Table 1.

Both views have received further support..Herrnstein (1966), Morse and Skinner (1958),

I Although they favored stimulus substitution as an ex-

planation for the majority of their data, Staddon and Sim-melhag also noted that it was "not adequate as a universalaccount" (1971, p. 16). They suggested that a number ofadditional principles of behavioral variation were re-quired to account for more complex behavior (see also

Staddon, 1983).

SUPERSTITION RECONSIDERED 281

plemental feeding in their home cages within1 hr after their daily sessions.

Table 2

Behavior categories, with subcategories indicating varia-tions coded within the major categories.

ApparatusThe experimental area was contained in a

large 81 by 76 by 45.7-cm sound-attenuatingchamber. The floor, back, and right walls ofthe experimental area were the sheet-metalwalls of the chaniber. In the initial studies,the left wall consisted of a standard two-keyLehigh Valley Electronics pigeon panel cen-tered in a 76-cm long piece of black plywood.In Experiments 4, 5, 7, 8, and 9, this wallwas covered by a flat gray 0.32-cm layer ofMasonite with an opening for the feeder. Thismodification made the surface of the wall moreuniform and controlled for any effects of thereflectiveness of the aluminum panel. Thefront wall of the experimental area was thedoor of the sound-attenuating chamber andcontained a double glass window, 61 by 30.5cm, through which the birds were observed.Lighting was provided in the chamber by twooverhead incandescent 60- W bulbs behind dif -

fusing gratings. Masking noise and coolingwere provided by two exhaust fans mounteddirectly over the lighting fixtures. The floorof the chamber was covered with hewspaper .

Procedure

Adaptation to the food hopper was accom-plished in 1 day by placing the pigeon in thechamber with the hopper raised and illumi-nated. As soon as the pigeon ate for 10 to 12 s,the hopper was lowered and raised again.Across the next 15 to 20 presentations, theduration of access to food was gradually re-duced to 5 sand the time between food pre-sentations was increased to 15 s. Each subjectreceived a total of about 30 presentations onthis day. Following feeder training the sub-jects typically received 20 days of training with40 hopper presentations per day on a FT 15-sschedule. Each hopper presentation lasted 3 s,except as noted.

The behavior of the pigeons was observeddirectly by one or two experimenters seatedin a dark room approximately 1.5 m from thechamber, and was coded into one of the cat-egories defined in Table 2. The coding cate-gories were developed by modifying the set ofcategories outlined by Staddon and Simmel-hag (1971) and Skinner (1948). The stylizeddrawings in Figure 1 show examples of typ-

PECKS-A discrete forward or sideways movement ofthe head in which the beak makes contact with asurface

PECKH-Peck inside or on the edge of the hopper

openingPECKW-Peck other wall areasPECKF-Peck floor, either with sideways brushing

motion or direct peck, most often related to picking

up spilled grainWALLD-Behavior directed to the wall nearest to or

containing the hopperBOB.W-Bob head in front of the hopper wall, either

side to side or up and down, usually with head upand neck elongated, but includes low bob on wall aswell

HI.BOB-A further subdivision of Bob. W used in laterstudies that included only bobbing directed towardthe top of the wall

LOBOB-A subdivision of Bob. W used in later studiesthat included only bobbing directed toward the bot-tom of the wall

WALKW-Bob and step back and forth in front ofwall

HUG.W~Walk wall while pushing breast against itSCRAW-Hug wall while scratching and climbing

against wall with feetSTAWY-Remain nearly motiol:!less (stand away), most

often near a front corner. Posture varies from uprightwith neck slightly extended to body at 45° with neckless extended

HOPPR-Oriented waiting within 8 cm of hopper withhead bobbing

BOB.H-Bob at wall hopperBOBFH-Bob at hopper in floorSQTFH-Squat over floor hopper with body horizon-

tal, usually turning in tight arcs around itORMI.D-Active waiting while oriented toward middle

of chamberBOB.F-Bob with head toward floor and not within 8

cm of a wallBOB.M-Bob toward middle, head up and not walking

LOCOM-Movement around the chamber in upright

postureWALKM-Walk around the chamberTURNS- Turn the body in at least quarter circle

EMOTN-Behavior apparently related to fear or escapeCROUC-Crouch, head low and not extended, tail highT AI.LF -Tail flick, a pronounced downward flick of

the tail feathersFLAPP-Repeated wing flapping, sometimes with re-

traction of feetGROOM-Preening behavior including wiping of face,

nibbling and stroking of feathers with bill, andscratching of body or face with foot

NOMOV-A motionless almost horizontal posture re-sembling roosting, with head pulled against body;usually occurred only in extinction


~

Table 3

Methodological differences between Skinner (1948) andStaddon & Simmelhag (1971).It

~~

..LLO ...L..'

Deprivation weight 75"10 ad liblntertrial interval 15 sFeeder duration 5 sSize of chamber ?Experience of Naive?

subjectsNumber of Few (1-2?)

sessionsType of Ad-lib

observation inspection

~

OOT"'_.~

Lmm ,

80"10 ad lib12 s2 sStandardNaive &

experiencedVery large

(>60)Systematic

categories

Fig. 1. Stylized drawings of typical postures in rela-tion to the hopper wall (here assumed to be on the left)for most of the general behavior categories. See Table 2for detailed descriptions.

ical postures for most of the major categoriesreported in this paper. A category of behaviorwas recorded when a click was produced every3 s within the 15-s interfood interval. Addi-tional notes were made on the scoring sheet tofacilitate interpretation and recall of the ses-sion. In addition, a sample of each bird's be-havior was filmed at asymptote. Unless other-wise noted, the data reported are averagesbased on the last 4 days of each experimentalcondition.

Four different primary observers were usedsingly in these experiments. The observerswere undergraduates who received eithercourse credit or pay for their work. Each ob-server read Skinner (1948) and selected partsof Staddon and Simmelhag (1971), and learnedthe list of behavior categories shown in Table2. One of the authors (G.A.L.) then sat witheach observer for 3 to 5 days while the ob-server coded each behavior pattern on a datasheet while calling it aloud. Immediate feed-back was given for appropriate and inappro-priate categorization. This form of guidedcoding continued until agreement on the ma-jor categories was close to 100%. Subsequenttests of interobserver agreement on the majorcategories listed in Table 2 have shown above90% agreement (the agreement for three re-cent pairs of observers was 92%, 92%, and93%). All coding was blind to the obtainedoutcome in the general sense that the observ-ers could expect only results similar either tothose of Staddon and Simmelhag (1971) or tothose of Skinner (1948). In addition, Exper-iment 2 was run specifically blind as to whichbirds had been pretrained to turn and whichto peck.

THE EMPIRICAL PHENOMENONOF SUPERSTITIOUS BEHA VIOR

EXPERIMENT 1The purpose of the first experiment was to

establish the nature of superstitious behaviorin our experimental conditions. An importantstep in this process was the selection of proce-dures from the slightly different methodologiesemployed by Skinner (1948) and Staddon andSimmelhag (1971) that are summarized inTable 3. In this experiment we chose the 15-sinterfood interval and 5-s hopper durationused by Skinner. We averaged the deprivationlevels by using a 77.5% body weight. We alsoran an intermediate number of sessions (20),and used a relatively large experimental areaso that the orientation of the birds' behaviorwith respect to the food source was relativelyunconstrained by adjacent walls. As notedabove, the behavior of the pigeons was clas-sified into the set of categories shown in Table2 and was recorded at 3-s intervals during theinterfood interval.

MethodThe subjects were 4 pigeons maintained and

housed as described in the General Methodsection. These birds had received prior auto-shaping training in standard pigeon boxes(Lehigh Valley). The apparatus, procedures,and recording techniques were those describedin General Method, except that each hopperpresentation lasted 5 s.

Results and DiscussionFigure 2 shows the temporal pattern of in-

terfood behavior plotted by 3-s intervals and


H18395-SEC~ e-m

* EMOTNA LOCOM~ ORMIO* PECKSO STAWY[!J WALLO

~.6 12

H1416,{)a 100 0000

H14165-SEC

~

H1B39

,11!.m.~.I!J-m.1!1.m-JIj.1!11!"

-.-PECKS..-6... WALLD--6- -LDCDt1-~ -STAWY-.-DRI11D

A. ft

Itt.~..~ 0 10 20

\8"'zIJ.J(.j SoQ:IJ.J~

" :2:lJ.JucrlJ.Ja..

50 50;0 '"i

rn

~.

6 12

000 0

H5327HS327 S-SEI 001,00 100

H6974

rn-em-m.i!J-em-m.Jn.Of--zUJu0::w0;.

So

0

0 6 12

SECONDS

50So

b-A-A-&.t.-A

t!;~~.~ ;..,.

10 20

DAYS

00

~

6 12

SECONOSDAYS

Fig. 3. Percentage of total behavior categories record-ed in Experiment 1 plotted across 2-day blocks. Note thatthese averages are taken across all the observation inter-vals.

Fig. 2. Percentage of behavior categories recorded inExperiment 1 across 3-s observation points during theinterfood interval. Data are means of all intervals for thelast 4 days of training. Note that only the four most fre-quently occurring categories for each bird are plotted.

after a relatively few trials when the behaviorwas more variable. However, Figure 3 showsthat wall-directed behavior emerged during thefirst block of two sessions. Wall-directed be-havior was simply .what pigeons did almostimmediately in this situation when access tofood was regularly presented independent ofresponding.

The second striking finding was that thetopographies we obtained differed from thosereported by either Staddon and Simmelhag(1971) or Skinner (1948) in important ways,though they resembled aspects of both. First,although behavior frequently appeared to bedivided into two classes that could be de-scribed as interim (behavior away from thehopper) and terminal (behavior near the hop-per), we never saw pecking of the magazinewall. Instead, the pigeons head-bobbed,stepped, and bumped, often throughout theentire interfood interval. On the other hand,although we saw elements of most of the to-pographies reported by Skinner, typically thesewere components of overall behavior directedaway from or toward the hopper wall. Forexample, although head-bobbing (one of thepatterns Skinner reported to be accidentallyreinforced) increased over trials, all the ani-mals did it in some form. Further, it was typ-ically directed along the wall above the hop-

averaged over all trials in Sessions 17 to 20.In this and subsequent plots, only the fourmost frequently occurring categories of behav-ior are shown for each bird. There are twostriking findings. The first is that all the birdsquickly exhibited almost identical behavior ,which could be divided into two broad classes,frequent circling away from the hopper im-mediately after it was lowered, followed orsuperseded by the predominant class of repet-itive movements along the hopper wall untilfood was presented. The forms of wall-di-rected behavior included neck up, head bob-bing, pendulum motions of the neck, slightlydrawn wings, stepping back and forth, bump-ing the panel with the breast, and scratchingthe wall with the feet. An alternative patternfrequently shown by Subject H5327 includedstanding nearly motionless near the wall andabout 20 to 30 cm to the side of the hopper

(stand-away).The overwhelming similarity of the behav-

ior patterns of the different birds makes veryunlikely Skinner's contention that accidentalcontingency is the cause of superstitious be-havior. Although there was variation in thebirds' behavior, it was definitely secondary tothe high level of similarity. It might be arguedthat Skinner's (1948) observations were made


per, and was accompanied by bumping,stepping, and neck and head motions..

In summary, the results resembled those ofStaddon and Simmelhag (1971) in showing acommon set of terminal behavior patterns, andoften a sequence of different types of behaviorwithin a trial.. However, the absence of peck-ing appeared to rule out the operation of thestimulus-substitution mechanism invoked byStaddon and Simmelhag to explain theirpecking results.. Similarly, the results resem-bled those of Skinner (1948) in showing in-creases in a variety of individual responsecomponents.. But the overall similarity andpatterning of the obtained behavior appearedto violate Skinner's argument that supersti-tious behavior was produced by accidental re-sponse-reward contingencies, and supportedeli.citation rather than operant response con-ditioning as the primary causal mechanism.

Although the consistency of these resultsstrongly suggests the importance of elicited,preorganized patterns in superstitious behav-ior, the data do not rule out a contribution ofresponse-reward contingencies, or the poten-tial importance of pecking. The next group ofexperiments analyzed more carefully the po-tential role of accidental response contingen-cies in producing the obtained behavior. Thesubsequent group of experiments explored thedeterminants of pecking in the superstition

paradigm.

THE ROLE OF RESPONSECONTINGENCIES

The experiments in this section exploredthe importance of response-food contingenciesin producing and controlling the behavior ob-tained in the superstition paradigm. Experi-ment 2 manipulated the initial probability 0fparticular responses by imposing prior re-sponse-contingent food deliveries. Experiment3 extinguished previously acquired supersti-tious behavior, and then examined the resultsof reacquisition. Experiments 4 and 5 assessedthe effects of explicit positive and negative re-sponse contingencies on wall-directed behav-ior.

responses by means of response-contingentfood delivery prior to exposure to the re-sponse-independent fixed-time schedule. Ourstrategy was to develop a response of initialhigh probability (either wall-pecking or turn-ing) to increase the likelihood of accidentaljuxtapositions of that response with food. Twobirds were trained to peck the magazine wall,and 2 birds were trained to turn in circles.The birds were then introduced to an FT 15-sschedule. During 3 days of response-contin-gent training, we gradually extended the con-tingent interfood interval to 15 s so that theanimal would less readily discriminate thesubsequent transition to a response-indepen-dent schedule (Herrnstein, 1966). It should benoted that naive birds were used in this ex-periment to evaluate the possibility that pre-vious autoshaped training had contributed tothe results of Experiment 1.

MethodThe subjects were 4 naive pigeons. Hous-

ing, feeding, apparatus, and initial pretrain-ing were the same as in Experiment 1. On thefirst day of training, 2 birds were randomlyassigned to a "pecking" condition and the oth-er 2 to a "turning" condition. Coincident withthe completion of hopper training, reinforce-ment of successive approximations was usedto shape either pecking at the magazine wallor turning in a circle. On the second day oftraining, the interfood interval was graduallyincreased to approximate an FI 15-s schedulewith 3-s hopper duration. On the third day oftraining, each subject received one full sessionof forty 3-s food deliveries on an FI 15 sched-ule. On the next day, the birds were trans-ferred to an FT 15 schedule with the samehopper duration; this procedure was contin-ued for 20 sessions. The observer was not in-formed of the subjects' pretraining.

ResultsFigure 4 shows the acquisition of supersti-

tious behavior plotted by two-session blocks.It can be seen that the level of the trainedbehavior (either pecking or turning), althoughinitially high, declined rapidly as wall-direct-ed behavior emerged. Figure 5 shows that atasymptote the behavior patterns of the birdswere similar and resembled those of the pi-geons in Experiment 1. The responses of

EXPERIMENT 2

In this experiment we modified the initialdistribution of behavior by briefly shaping two


flO92 "URN HO921 TURN Hl428 PECK

d EMOTN !:. LOCOM

~ ORMIO

* PECKS

~ STAWY

1!1 WALLO

1004 ..1~O 1.00 00

1-ZwU 50~W0:.

50 50

~0 0 0

60

H6B46 TURN H6BO6 PEC H6846 TURNiou 100 100H6BO6 PECK

~f-zwU 56a:W0.

,ID-m-m-eJ-{!1.m.{!1.m

xI~so

~~ oO 6 12 0 6 12

SECONOS SECONOS

Fig.5. Percentage of behavior categories recorded inExperiment 2 across 3-s observation points during theinterfood interval, during the final 4 days of the FT 15-sschedule of food delivery.

f-zlJ.JU 50~lJ.Ja.

~90

accidental response contingencies were at workin our situation, the present birds should showdifferent responses during reacquisition.

Method

The subjects were 4 experienced pigeons.Housing, feeding, apparatus, and initial pro-cedures were identical to the previous exper-iment. Acquisition for these birds is reportedin Figure 11 (Experiment 6b). Following ac-quisition, no hopper presentations occurred for12 sessions. Food was then presented again onan FT 15 schedule for 8 days. The hopperduriition used in this condition was 2 s.

Results

The data in Figure 6 indicate quite thor-ough extinction after 12 days of no reward.The birds remained motionless in a single po-sition during nearly the entire session. How-ever, Figure 7 shows that when food was re-introduced, the birds rapidly resumed thetypical patterns of wall-directed behavior (seealso Figure 11). Thus, on the basis of Skin-ner's argument, accidental contingencies of re-sponse and reward appear to have had littleto do with the particular behavior that oc-curred. The alteration of behavior Skinner(1948) reported may have reflected the devel-

pecking the hopper wall and turning, despitetheir initial high probabilities, were super-seded by a basic set of wall-directed patterns.However, there appeared to be an indirect ef-fect of prior response-contingent training onthe components of superstitious behavior thatemerged. The birds trained initially to turnrather than peck showed a greater variety oflocomotor behavior and more of the stand-away form of wall-directed behavior. Evenwith these minor differences, the outcomeagain supports the view that superstitious be-havior is basically elicited rather than rein-forced, although the elicitation does not fit theclassic stimulus-substitution model.

EXPERIMENT 3

In this experiment birds previously trainedon an FT 15 schedule were thoroughly ex-posed to an extinction procedure and thenplaced again on the FT 15 schedule. Skinner(1948) reported using this procedure with asingle bird and finding different responses thesecond time. He cited this result as proof ofthe effects of accidental contingencies betweenresponses and food under the FT schedule.When food was reintroduced, simply by chancethe contingencies favored a response other thanthe one just extinguished. It follows that if

UJ~ ~B1, ~m-ID.'e3-JIJ-qJ"- Q. b-&; .~h-~A

),'0 --

0 10 20

J.):'-~ -

j ~~j!}'-I1~b-"-~-6.b-t; -~ A_A.A-6-b-t;

Wi~".;:i, 00 10 20 0 10 20

DAYS DAYS

Fig. 4. Percentage of total behavior categories recordedin Experiment 2 plotted across 2-day blocks. The solidpoints to the left of the first block of days in each graphrepresent the level of performance of the pretrained be-havior on the last day of FI 15-s training.


R2526 RECOND

~O HOPPRt. LOCOMX NOMOV* PECKS(!J WALLO

4~

R1423 EXT 00.,l~O 100R2526 EXl

~&-&X~ EMOTN/!j HOPPRA LOCOMX NOMOV* PECKSI!] WALLD

~

O 6 12

~zwU iSO

ffi~0-

'""""2:LlJuCt:

i:1J

~1

50

0 00 6 ,2 0

R3700 EXT

&-g gg 9

R3B77 EXTg~~~g!0 IOO

~,50

t-7'

So\.,.)~

~~

0 i;.. , .~ 01.O 6 12 O 6 12

SECONDS SECONDS

Fig. 6. Percentage of behavior categories recordedduring the final four sessions of extinction in Experiment3, across 3-s observation points corresponding to periodswithin the interfood interval during the prior and subse-quent FT 15-s schedule.

opment of alternative forms of wall-directedbehavior or a decrease in locomotor patternsas a function of repeated extinction and re-

conditioning.

EXPERIMENT 4

Although the previous experiments stronglyindicate that behavior in the superstition par-adigm is based on preorganized rather thannewly shaped response patterns, nonetheless,there still may be a role for response-rewardcontingencies in maintaining and modifyingsuperstitious behavior. In the next two exper-iments we attempted to modify behavior bymaking food presentations contingent on re-sponding. In the present experiment, access tofood was contingent upon experienced birds'wall-directed behavior (defined as remainingwithin 8 cm of the panel and within 8 cm ofeither side of the hopper). In the first operantcondition, food was presented immediatelywhen the bird had remained in this area for3 s. Thus, birds were able to decrease the timeto food (and increase the frequency of fooddeliveries) by returning to the hopper morequickly, or by never leaving the area. In thesecond operant condition, frequency was heldconstant by imposing the additional rule thatonly one food delivery could be produced ineach 15-s interval. In both conditions, if the

wall-directed behavior of the birds on the FT15 schedule was being maintained by acciden-tal response contingencies, the addition of ac-tual response contingencies should produce arapid increase in the relative frequency andintensity of wall-directed behavior because itnow more consistently led to food. However ,to the extent that components of superstitiousbehavior were elicited, the behavior shouldprove less amenable to easy modification byresponse-food contingencies.

MethodThe subjects were 4 experienced pigeons

that had demonstrated moderate to high levelsof wall-directed behavior under 20 days of FT15 delivery of food. In this and all of the fol-lowing studies (except Experiment 6), the ap-paratus was modified so that the hopper wallthat held the feeder was covered by a flat gray0.32-cm layer of Masonite. This change madethe hopper wall more uniform, thus control-ing for the possibility that wall-directed be-havior was related to the reflectiveness of thealuminum panel. The experiment had fourphases: baseline (FT 15), fixed-duration 3 s,contiguity-promoting 15 s, and baseline (FT15). Each phase was maintained until re-sponding stabilized and a minimum of 6 dayspassed.


As discussed above, the fixed-duration 3-s(FD 3) schedule rewarded any approach tothe wall that was maintained for a minimumof 3 s. The contiguity-promoting (CP) pro-cedure was an alternative FD 3/FT 15 sched-ule that provided one 3-s hopper presentationduring each 18-s cycle. The first time the pi-geon maintained wall-directed behavior for aminimum of 3 s during each 18-s cycle, grainwas presented immediately; otherwise, grainwas presented during the last 3 s of the cycle.Thus, this schedule promoted contiguity be-tween wall-directed behavior and food deliv-ery while maintaining a constant rate of fooddelivery.

P3377

0-z"'u0:

~

Fig. 8. Percentage of wall-directed behavior obtainedunder three reinforcement contingencies in Experiment 4.Bars show standard errors of the mean. Data are meansbased on the final four sessions of the condition.

prevented a marked effect of the response con-tingencies in one bird; however, this does notaccount for the maintenance of response to-pography, nor for the failure of the other birds'rewarded behavior to increase markedly or

consistently.

EXPERIMENT 5

In this experiment we trained naive ani-mals on an omission procedure contingentupon active wall-directed behavior. Response-based omission of reward has been a classicmethod in testing for elicited behavior (Shef-field, 1965; Williams & Williams, 1969). Tothe extent that wall-directed behavior or re-lated interim components emerge and con-tinue under an omission contingency, anelicited basis for superstitious behavior isbetter supported. To the extent that wall-directed behavior becomes less frequent, a rolefor response contingencies in superstitious be-havior is better supported. We note that thesearguments are not entirely conclusive as tocausation of responding (e.g., jenkins, 1977),but we consider the results of the omissionprocedure potentially instructive.

Results

Figure 8 shows the mean percentage ofwall-directed behavior maintained at asymp-tote under each schedule condition (the twoFT 15 baseline conditions produced similarresults and were averaged). Under the fixed-duration procedure, 3 of the subjects actuallyshowed a slight decrease in percentage of wall-directed behavior. This decrease appeared tooccur because, although the contingencyshortened the intertrial interval, the pigeonsstill walked away from the magazine aftereating. Thus, wall-directed behavior made upa smaller portion of the interfood interval. Inthe case of a subject that previously showedno locomotor behavior (P1186), there was aslight increase in percentage of wall-directedbehavior. However, when the length of theintertrial interval was controlled by using thecontiguity-promoting schedule, there was noconsistent change from the percentage of wall-directed behavior under the fixed-time con-dition.

The basic topography of the behavior pat-terns observed under all conditions remainedsimilar. However, for 2 of the birds the meanonset of wall-directed behavior occurredslightly earlier in the interval under the re-sponse-contingent procedures. These birdssimply shortened the duration of the walk-

away pattern.In sum, explicit response contingencies

modified the superstitious behavior of 3 of thebirds in some way, but not to the extent orwith the consistency one would expect if su-perstitious behavior were maintained primar-ily by accidental reinforcing consequences. Itmight be argued that a ceiling on responding

Method

The subjects were 4 naive pigeons. Thehousing, apparatus, and preliminary proce-dures were identical to those of Experiment

WILLIAM TIMBERLAKE and GARY A. LUGAS288

P1990 P2003

I

;

~ kU0 10

4. Food was presented on an FT 15 scheduleso long as the bird was not within an areaextending 8 cm from the front wall and 8 cmon either side of the hopper during any of the3-s periods in the interfood interval. An ex-ception was made during the first 3-s intervalto allow the birds an opportunity to leave thehopper area after the hopper was lowered.

Results

The percentages of wall-directed behaviorrecorded across days for each subject are shownin Figure 9. Despite the presence of an omis-sion contingency, all 4 pigeons initially devel-oped a strong tendency to step, bump, andhead-bob along the feeder wall. For one Sub-ject, P1990, this wall-directed pattern wasquickly replaced by the stand-away pattern.However, another subject, P2003, continuedwall-directed stepping and bobbing despite theomission procedure. For the other 2 pigeons,wall-directed responding decreased over days,and the wall-directed patterns were replacedmainly by increased circling and bobbing di-rected toward the middle of the chamber.Thus, a negative response contingency de-creased wall-directed behavior in most birds,but the initial emergence and continued oc-currence of active wall-directed behavior still

,0

C\'1-Zwut)::w().. ~

Ccc

0 10

o~.~0 10

P200B P2026

~

AzwuCYw0-.

m

v ~

0 10

DAYS DAYS

Fig. 9. Percentage of wall-directed behavior obtainedin Experiment 5 during successive sessions of an FT 15-sschedule that omitted food deliveries when wall-directedbehavior occurred.

suggest control by elicitation. Further, that thebirds engaged in behavior patterns already ob-served under response-independent food de-livery (e.g., stand away and turn) suggests thepossibility that omission had its effect by al-tering the forms of elicited food-related be-havior, rather than by rewarding some arbi-trary new responses.

DISCUSSION OF EXPERIMENTS2 TO 5

These experiments support the view thatbehavior in the superstition paradigm does notdepend primarily on accidental response con-tingencies, nor is it dominated by pecking, asexpected from a stimulus-substitution ac-count. Apparently the behavior depends on theelicitation of preorganized behavior patternsby the frequent presentation of food. Limitedforms of wall-directed and interim behavioroccured whether they were previously extin-guished, whether pecking or turning had pre-viously been reinforced, whether food deliv-eries were contingent upon remaining near thehopper, or whether food was contingent uponstaying away from the hopper .

However, there were effects of responsecontingencies on behavior. Although pretrain-ing birds to peck or turn did not produce peck-ing or turning as terminal behavior patterns,the pretraining did modify the subsequent dis-tribution of responding. Also, making food de-liveries contingent upon the birds' standingnear the hopper wall affected some aspect ofthe wall-directed behavior of 3 birds. Finally,an omission contingency for the birds' re-maining near the wall markedly decreased theinitial level of wall-directed behavior in 1 ofthe birds and produced a decreasing trend inwall-directed behavior in 2 others.

An interesting discovery in the omission ex-periment was that when food was omitted forstanding near the wall, the birds did not en-gage in random alternatives. Instead, 1 birdengaged in a stand-away pattern, while theother 2 circled; Both of these types of behaviorhad been observed previously. Stand-away wastypically a much less probable alternative toactive wall-directed behavior. Circling awaywas a predominant pattern following food andpreceding wall-directed behavior. Thus, evenwhen the omission contingency was successfulin reducing wall-directed behavior, the nature

50 j

~~~


ment 1 we used birds previously trained onan autoshaping procedure and this previousexperience did not result in pecking duringour experiment. Whether different or addi-tional schedule experience would have an ef-fect must remain conjecture. Second, there maybe strain or gender differences in likelihood ofproducing pecking (e.g., Fenner, 1980). Third,the shorter, 12-s interfood intervals used byStaddon and Simmelhag increased the fre-quency of food deliveries and thus may haveincreased the likelihood of pecking (see Innis,Reberg, Mann, Jacobson, & Turton, 1983;Innis, Simmelhag-Grant, & Staddon, 1983).Fourth, the use of a 2-s hopper presentationmay have forced greater attention to the mag-azine area to avoid missing the food, and thismay have indirectly promoted pecking. Fifth,most of the experiments that reported peckingran for great lengths of time, often 4 to 6months, instead of the 20 days we typicallyran. Pecking may be late-emerging behavior .Sixth, some investigators have included in theirpecking category head-bobbing movementsthat did no~ actually involve beak contacts.Head bobbing is pigeons' normal pattern ofvisual scanning (they have very limited eyemovement), and mayor may not lead to peck-ing (see Zweers, 1982). It is important,though, to stress that counting head bobs aspecks cannot be the entire story, because thereis no question that considerable actual peckinghas been obtained. Finally, it might seem thatthe occurrence of pecking could have beenunderestimated in the present data because abrief peck would be likely to be missed undera 3-s sampling scheme. However, we removedthis possibility after the first experiment byasking the observers to note any pecking in aninterval, regardless of whether it occurred atthe sample point. The experiments reportedbelow explored the contributions of many ofthe above circumstances to the elicitation ofpecking.

of the replacement patterns still suggested theexistence of preorganized species-typical pat-terns elicited by the repeated delivery of food.Thus, the results can be used to argue for theexistence of a limited hierarchy of behaviorpatterns related to feeding: If one is punished,another of predictable topography will take itsplace (cf. Murray, 1973, cited in Williams,1981).

A final issue was the apparent contradictionof previous data showing that pigeons thatwere transferred from response-contingent toresponse-independent schedules continued topeck (e.g., Herrnstein, 1966). In Experiment2, 2 pigeons stopped pecking and engaged inwall-directed behavior when exposed to thistransition. We suggest: (1) Extensive trainingmay markedly increase the dominance ofpecking as an appetitive behavior pattern; (2)extensive responding to a lighted key may in-crease its salience as a target; (3) direct ob-servation of birds in previous studies mighthave revealed the development of considerablewall-directed behavior accompanying thepecking that was recorded.

Having established the likelihood that be-havior in the superstition paradigm is pri-marily elicited rather than controlled by ac-cidental response contingencies, it is importantto begin to analyze what controls the complexset of wall-directed behavior patterns. The ex-periments in the next section attempted to ex-plore the reasons for differences between theset of elicited behavior patterns we obtainedand the apparently simpler pattern of peckingthe hopper wall reported by Staddon and Sim-melhag (1971) and others.

FACTORS THAT INFLUENCETHE ELICIT A TION OF

SUPERSTITIOUS PECKING

The next several experiments were an at-tempt to discover what variables might ac-count for the absence of pecking of the wallnear the hopper (e.g., Staddon & Simmelhag,1971). A careful reading of the many articlesreporting terminal pecking revealed severalpossibilities. First, birds that had previous ex-perience in working on schedules were morelikely to peck than were naive birds. Onlyabout 50% of naive birds pecked consistently.It should be noted, however, that in Experi-

EXPERIMENTS 6A AND 68

Experiments 6a and 6b tested the possibil-ity that a short, 2-s hopper presentation wouldresult in terminal pecking. A short hopper du-ration might cause the birds to remain closeto the hopper and thus increase the likelihoodof their pecking the wall. In Experiment 6abirds with previous experience with aS-s


100100R25262-SEC

~~ HOPPR6 LOCOMX NOMOV* PECKS

~

6 12

t-zwU 50~W~

So

0006 120

R38772-SEC

,.--'"-.,JOO1°°

1 R37002-SEC

~~

WU So~lJ.J0..

"'

:x50

A[9:"

~..~O 6 12 O 6 12

SECONDS SECONDS

Fig. 11. Percentage of behavior categories recordedfor the naive birds in Experiment 6, plotted across 3-sobservation points during the interfood interval. Data aremeans based on the final four sessions of the experiment.

0 0

hopper duration on an FT 15-s schedule wereshifted to a 2-s duration. Experiment 6b con-tinued this investigation by exposing naivebirds to an FT 15-s schedule with a 2-s du-ration from the beginning of training.

MethodThe subjects for Experiment 6a were the

birds used in Experiment 1. The subjects forExperiment 6b were 4 naive birds. Housing,feeding, apparatus, and basic procedures werethe same as in Experiment 1. Each set of an-imals received 40 2-s hopper presentationseach day. The experienced birds received 10daily sessions of this procedure; the naive birdsreceived 20 sessions.

ResultsThe results for both sets of birds were sim-

ilar and are shown in Figures 10 and 11. Thesubjects showed little behavior that resembledpecking of the hopper wall. Instead, theyshowed the now familiar sequence of somelocomotor behavior after obtaining food, fol-lowed by wall-directed behavior until the nextdelivery of food. Again, the wall-directed be-havior was predominantly wall "hugging,"including bumping, stepping, pendulum mo-tions of the head and neck, and bobbing of thehead. Only a few birds showed the passive

stand-away form, and only one, H5327,showed it more than occasionally. Thus, the2-s hopper duration did not by itself increasethe likelihood of pecking.

EXPERIMENT 7

Experiment 7 explored the possibility thatdifferences in strains of birds may account fordifferences in the elicitation of pecking thehopper wall as a type of terminal behavior .For example, Fenner (1980) reported very lit-tle pecking in White Carneaux pigeons in asuperstition paradigm, but more pecking byWhite King and Silver King pigeons. Thus,Experiment 7 used female White Kings assubjects in an attempt to increase the likeli-hood of pecking. Innis, Simmelhag-Grant, andStaddon (1983) reported that pecking almostnever occurred at intervals greater than 12 s,although wall-directed behavior continued tooccur at intervals up to 300 s. Interfood in-tervals as short as 6 s appeared to increase thelikelihood of pecking (see also Innis, Reberg,Mann, Jacobson, & Turton, 1983). Thus, ina second phase of Experiment 7 we decreasedthe interfood interval from 15 to 9 s.

MethodThe subjects were 4 female White King

pigeons. Housing, feeding, apparatus, andpretraining procedures were similar to Ex-


100 lOp OQc 100K7922 FT 9

Jt-"*f-zwU 50~WQ.

*.,( (!) !

~

PECKSWALLD50

~zwU So0:

~

50

0 0 b 06 12

KB729 FT 15 K8949 FT 15 KB729 FT ~ K8949 FT 9100 100

-m---m1-ZwU So~WQ.

~~

50

0

~ -l "--6~

0-:' ., =IO 6 O 6

SECONDS SECONDS

Fig. 13. Percentage of behavior categories for theWhite King pigeons in Experiment 7, across 3-5 obser-vation points during the interfood interval under an FT9-5 schedule. Data are means based on the final foursessions of the condition. .

periment 1, except for an initial exposure toan FT 15-s schedule with a 3-s hopper pre-sentation through a hole in the floor of thechamber, 20.3 cm from the plywood wall.Following 16 sessions of this condition, thebirds were switched to the usual wall-mount-ed feeder for the 8 sessions reported here. Fol-lowing this procedure the schedule waschanged to an FT 9-s schedule for an addi-tional 8 sessions. Data are reported from thelast 4 days of each of the two conditions withthe hopper in the wall.

ResultsAs shown in Figure 12, for one of the birds

(K7918) the results were very similar to thoseobtained before. Behavior was largely dividedinto a moderately probable leaving of the hop-per area, followed by wall-directed behaviorthat continued until the delivery of food.However, 2 birds showed considerably morelocomotor behavior, and the fourth, K7922,developed a persistent pattern of pecking thewall below the hopper. Pecking in this birdemerged when the feeder was in the floor andwas directed at the rim of the floor-hopperaperture. When food was delivered from thewall, the pecking moved to the area below thathopper. Changing the interfood interval to 9 s,as shown in Figure 13, increased the percent-

age of pecking for K7922 and the percentageof wall-directed behavior for the other birds.

Although not definitive, our results are con-sistent with Fenner's (1980) demonstration ofsmall strain differences in the amount of super-stitious pecking, and suggest that a reasonpecking has been reported more frequently insome previous experiments is that these ex-perimenters typically used strains other thanCarneaux, such as Silver King or White King.Further, the results indicate that a 9-s inter-food interval produces a greater proportion ofwall-directed behavior of all kinds than doesa 15-s interval. These results may have beenhighlighted in the present data because theprior training with the hopper in the floorappeared to produce an unusually largeamount of locomotor and orient-to-the-middlebehavior. However, this behavior is unlikelyto represent a strain difference because simi-lar behavior was obtained using a floor hop-per with Carneaux pigeons in Experiment 9.

EXPERIMENT 8

The final possible factor affecting peckingthat we explored was the large number of ses-sions typically used in experiments reportingpecking. Although we had previously run birdsin Experiments 1 and 6a for a combined totalof 40 days without obtaining pecking, none-


dulum movements, and head bobbing. Thebirds began to peck only after assuming a morehorizontal posture for a number of days whilecontinuing to head bob in front of the panel(low bob). The subject that began peckingearliest, P1339, also showed the earliest emer-gence of this low-bob posture. Stepping, pen-dulum motions, and bumping decreased at thispoint. After a variable number of days in thislow-bob posture, the birds began to peck thehopper wall, typically within several inches ofthe top of the hopper opening, but not insideit.

PO276--PECKS---e3--. HIBOB

'~1-,-zwu~

" .

SDf.,;

0 50

P6361

[!J-~,.pJ-m.~~m.!!1-3 ~

~ ::;::;:; A-& ",,-K 'tr-K

0 50

0

DAYS OAYS

Fig. 14. Percentage of total behavior categories re-

corded in Experiment 8 plotted across 5-day blocks.

DISCUSSION OF EXPERIMENTS6 TO 8

These results supported our previous con-clusion that a relatively small and consistentnumber of interim and terminal behavior pat-terns dominate the superstition situation. Un-der nearly all conditions, the pigeons engagedprimarily in wall-directed behavior of the typ-ical form. It remains difficult to argue that theoverall reliability of this behavior arises fromaccidental response-reward contingencies. Al-though there were variations in response to-pography, these variations occurred in a con-text of similarity that suggests that componentsof a preorganized set of behavior patterns isreadily elicited in the present stimulus cir-cumstances by the periodic presentation offood.

We were able to obtain a small amount ofpecking of the sort reported by Staddon andSimmelhag (1971) and others, but it requiredusing more sessions or a different strain ofpigeons. Even under these conditions, peckingwas not common, and it clearly emerged fromthe context of general wall-directed behavior.It should also be noted that one of the Car-neaux piegons that pecked was unique in thatshe pecked before Trial 20, something we hadobserved in no prior bird of this strain andgender. That we found pecking so difficult toobtain speaks against accounting for supersti-tious behavior in terms of a stimulus-substi-tution mechanism of elicitation (cited by Stad-don & Simmelhag, 1971). It speaks for theimportance of a more general view of the pat-terns of appetitive behavior available for con-ditioning (Timberlake, 1983b; see also Stad-don, 1983; Staddon & Simmelhag, 1971).

Our initial inability to obtain pecking maybe seen as fortunate in that it called attention

theless, experiments in which pecking is typ.ically obtained use an even greater number oftrials and no change in conditions. In this ex-periment we observed 4 naive female WhiteCarneaux pigeons for a total of 50 sessions.

Method

The subjects were 4 naive pigeons. Thehousing, feeding, apparatus, and pretrainingwere similar to Experiment 1. The birds wereexposed to 50 sessions with an FT 15-s sched-ule with a 3-s hopper duration.

Results

As shown in Figure 14, the results supportthe importance of extent of training as a vari-able contributing to pecking. Two birds even-tually pecked on approximately 20% of thesample intervals. Subject P1339 began peck-ing the hopper wall frequently around Session20, and Subject P1296 around Session 45.However, it is important to note that even forthese 2 birds, pecking was not the dominantpattern of behavior. They engaged in non-pecking wall.directed behavior more often than

they pecked.An interesting aspect of the results was the

occurrence of characteristic sequences of to.pographies that preceded the development ofpecking. Both birds began with the compo-nents of typical wall-directed behavior, bodyand head erect, bumping, stepping, neck pen-


to the reliable matrix of behavior that pre-cedes food delivery without allowing it to bedismissed as part of an immediate ingestivesequence. Inspection of the reports by Staddonand Simmelhag (1971), Reberg, Innis, andtheir co-workers, and especially Innis, Re-berg, Mann, Jacobson, and Turton (1983),and Innis, Simmelhag-Grant, and Staddon(1983) shows results comparable to our wall-directed behavior. This pattern was simplydeemphasized in their reports in favor of thefocus on pecking. In our estimation the largermatrix of appetitive behavior patterns is morelikely to hold the key to explaining supersti-tious behavior than an account focused on

pecking.

APPETITIVE STRUCTURE ANDSUPERSTITIOUS BEHA VIOR

EXPERIMENT 9

The final experiment to be reported herebriefly explored some ramifications of the hy-pothesis that wall-directed behavior consistsof elicited components of species-typical ap-petitive behavior ordinarily related to obtain-ing food. This hypothesis stems from the as-sumption that learned behavior is based onspecies-typical units (modules) of appetitiveresponse tendencies and stimulus sensitivitiesorganized within larger functional system re-lated to survival (Timberlake, 1983a, 1983b,1983c; Timberlake, Wahl, & King, 1982; seealso Baerends, 1976; Hogan & Roper, 1978;Hollis, 1982; Scott, 1958; Tinbergen, 1951).The emergence of common behavior patternsin the superstition paradigm suggests that thesuperstition procedure contacts such a set ofappetitive modules related to food, and thatthese modules are expressed in the patterns ofcircling away and wall-directed behavior .

Although it is not yet clear how wall-di-rected behavior fits into the pigeon's naturalrepertoire of obtaining food, the circling awayseems closely related to components of groundforaging. Observation of ground-feeding pi-geons shows that pigeons typically move awayfrom a location where they have found food(personal observation; see also Murton, 1965).The appetitive-structure view of conditioningcan be tested indirectly in these circumstancesbecause it predicts that behavior produced byresponse-independent procedures should be

influenced by ecologically relevant stimulusconditions {Timberlake, 1983b; Timberlakeet al., 1982). Thus, we should be able tochange the stimulus conditions to support morereadily interpretable forms of food-gettingbehavior. To this end the final experiment in-volved moving the hopper away from its lo-cation in the wall and into the floor at thecenter of the chamber, thus providing aground-foraging context that ecologically ismore typical for the adult pigeon.

MethodThe subjects were 4 naive pigeons. The

chamber was identical to that described in Ex-periment 1 except that a feeder was installedbeneath the wooden floor in the center of thearea, 38.1 cm from each wall. The feeder ap-erture was a funnel-shaped opening approx-imately 5 cm in diameter at the top and ta-pering to a 1.6-cm opening about 3 cm belowthe surface of the floor. When operated, thefeeder solenoid raised a food tray flush withthe bottom of the funnel opening. When raised,the food was clearly visible. No additionalhopper light was presented. As in the previousstudies, the floor of the chamber was coveredwith newspaper. A 5-cm hole was cut in thepaper to accommodate the hopper opening.The pigeons were trained for 12 sessions un-der a FT 15-s schedule with a 3-s hopperduration.

Results and Discus."ionAs shown in Figure 15, the location of the

feeder had a marked effect on the patterns ofbehavior generated by the periodic feedingschedule. Two subjects, D3312 and D33 77,developed a predominant pattern of locomo-tion and head bobbing directed toward thecenter of the chamber. Following the offset offood, each subject began to walk about thecenter of the chamber, head oriented towardthe floor, in the circling pattern typical of pi-geons foraging for scattered food on theground. The main difference between behav-ior patterns of these 2 birds was the extent ofcircling while bobbing toward the floor. Incontrast, Subject D4151 initially circled awayfrom the feeder and then spent most of theremainder of the interval engaged in steppingand bobbing toward the side wall, while Sub-ject Dl186 showed a mixture of wall-directedbobbing and circling.


tOO 00, 033\2

* PECKS(!) STAWY~ WALLD

~

\

.so~

)(

~~.~

6 12

0 1o03377 04151

100101

~rzll,Ju~~

~

6 12

SEt{)NUS

(!"50 50

1oo

~

~

6 12

SECONDS

Fig. 15. Percentage of behavior categories recordedacross 3-s observation points during the interfood intervalin Experiment 9 when the food hopper was located in thecenter of the floor. Data are means based on the final foursessions of the experiment.

Thus, the patterns of behavior that ap-peared when food was presented in the centerof the chamber contained distinct nonarbi-trary sets of locomotor and visual search pat-terns clearly related to foraging for groundsources of food. However, 2 birds also showedwall-directed behavior. Although these resultsdo not illuminate further the precise function-al role of the wall-directed behavior observedin the previous experiments, they suggest thatsomething about the wall itself supports step-ping, bobbing, and bumping in the context offood, whether or not food is located in it.

dividual styles congruent with Skinner's re-port. Some birds scratched more than others,some swung their necks more, some head-bobbed more intensely. But all showed mostof the components we have identified as wall-directed behavior, and patterns typically oc-curred along the wall containing the hopper .The large experimental chamber we usedprobably made the focus on the hopper wallmore obvious. In a smaller chamber, or in onewith the hopper in a corner (e.g., Innis, Sim-melhag-Grant, & Staddon, 1983), the orien-tation of these behavior patterns might be morevaried and/ or appear less directed.

Other reasons for our differences withSkinner (1948) are probably related to his fo-cus on the diversity rather than the common-alities of behavior patterns. First, he reportedany increase in a single response regardless ofits temporal location in the interval. For ex-ample, circling, which typically occurs earlyin the interfood interval, was cited by Skinneras "superstitious" behavior on a par with headbobbing, a pattern that occurs most frequentlyat the end of the interfood interval. Second, itis probable that the majority of Skinner's ob-servations were taken very early in trainingwhen diversity in behavior may have beenmore apparent (though we have no strong em-pirical evidence that this diversity persists forany period of time). Finally, it may be thatbecause Skinner did not systematically cate-gorize the birds' behavior, he was drawn toemphasize increases in the frequency of indi-vidual motor acts that fit his conception ofcausation by response-reward contingencies,rather than noting the larger organization ofpredictable behavior change.

Our results clearly support Staddon andSimmelhag's (1971) assumption that in a fun-damental way behavior in the superstitionparadigm is elicited by the periodic presen-tation of food. Response contingencies, whilepotentially affecting behavior, appear unlikelyto have been the primary basis of behaviorchange. Preliminary procedures that in-creased the a priori probability of pecking orturning did not produce more pecking or turn-ing in the behavior that subsequently emerged.When wall-directed behavior was explicitlyrewarded or punished, similar behavior con-tinued to occur. When other behavior patternsdid emerge under the omission contingency,they consisted of response forms previously~

GENERAL DISCUSSION

Chance Contingency versus Elicitation

On the basis of these experiments, we con-clude that Skinner's explanation for the oc-currence of superstitious behavior on shortfixed-time schedules is incorrect. Nearly all ofour birds showed a common set of wall-di-rected behavior, including bumping, scratch-ing, head bobbing, and neck movements infront of the hopper. Skinner ( 1948) reportedalmost all these patterns but his descriptiondid not capture their uniformity of topogra-phy, relatedness, or location. There is no doubtthat the present birds showed recognizable in-

01186

~ EMOTNto LOCOM~ ORM10

~.I. I.

6 12


by a truly remarkable commonality of sensi-tivities to infrequent accidental contingenciesthat somehow resist the effects of much moreexplicit programmed contingencies (Experi-ments 4 and 5). Both explanations point to ourconclusions that superstitious behavior comesfrom a specific evolved system selected to pro-duce species-typical appetitive behavior pat-terns in the presence of food.

Finally, some readers may agree that thebehavior we obtained in the superstition par-adigm is fundamentally elicited, but argue thatphenomena such as the persistence of keypecking in birds transferred from response-dependent to response-independent schedules(e.g., Herrnstein, 1966) still point to the effectof "superstitious" contingencies on behavior .It has been argued in such cases that the con-tinued "accidental" juxtaposition of key peck-ing and food maintains pecking in the absenceof an actual contingency. However, our datafrom Experiment 2 suggest that such acciden-tal contingencies do not work that simply. Ap-parently there is something unique abouttraining of pigeons' key pecking that producespersistence of this response under fixed-timeschedules. We think it likely that these resultsoccur because training a pigeon to peck alighted key produces an integrated responserelated to natural food-getting behavior. Oncetrained, key pecking is readily elicited by thecombination of frequent food presentation anda salient target. In the absence of changes inthe supporting stimuli or the frequency of foodpresentation that more strongly support otherspecies-typical forms of food-getting behavior ,key pecking should remain a dominant re-sponse.

PeckingOne concern regarding the present results,

given those of Staddon and Simmelhag (1971)and others, is the very limited amount of peck-ing we obtained. Although an elicitation anal-ysis of superstitious behavior does not requirethat pecking occur (Staddon, 1983; Staddon& Simmelhag, 1971), any appeal to classicstimulus-substitution processes to explain su-perstitious behavior argues that it should. Inthe present experiment, pecking did not occurwith short hopper durations, with short inter-food times, or even with food deliveries madecontingent upon the birds' remaining near thewall. We were able to obtain some pecking by

observed in this paradigm, rather than arbi-trary new patterns. In general, it appears thatwhat has been called superstitious behavior inthe pigeon actually represents expression ofpreorganized response patterns elicited byperiodic delivery of food.

However, it should be emphasized againthat the final form of superstitious behaviorcan be modified by previous stimulus condi-tions and response-reward contingencies. Inthe present experiments the components of su-perstitious behavior were seen to vary withpretraining (Experiment 2), with positive andnegative response contingencies (Experiments4 and 5), and with previous experience (Ex-periments 7 and 8). Because of these effects itmight be tempting to argue that response con-tingencies, although not producing initial wall-directed behavior, are responsible for itsmaintenance. While one cannot disprove thiscontention with certainty, there are elementsof this position that are not attractive.

If wall-directed behavior is originally elic-ited as a species-typical appetitive pattern andthen is shaped and maintained by accidentalresponse contingencies, behavior patternsshould develop in two stages. Behavior in theinitial elicited stage should be common acrossall subjects, while behavior in the subsequentreinforced stage should show large, unpre-dictable variation across subjects as accidentalcontingencies begin to modify the patterns.Although we cannot provide a precise evalu-ation of this prediction, it is not congruentwith our impressions. If anything, the behav-ioral topography initially is more variable andsubsequently settles into the typical wall-di-rected patterns. The failure of a continuingpositive contingency for remaining near thehopper wall to affect behavior systematicallyalso speaks against maintenance of wall-di-rected behavior by response contingencies.

One might still argue for a fundamentalrole of response contingencies if one assumedthat superstitious behavior were based on acommon precise response-reward contingencythat nearly all birds were programmed to "fallinto" under a variety of different procedures.However, this argument, in addition to beinguntestable, makes essentially the same pointwe wish to make with an elicitationappproach. Basically, superstitious behaviormust be explained either by elicitation throughpairing of food with the passage of time, or

WILLIAM TIMBERLAKE and GARY A. LUGAS296

selecting a different strain of bird and bytraining Carneaux pigeons over a longer pe-riod of time. In the Carneaux there appearedto be a particular acquisition sequence. Peck-ing emerged after the bird developed a con-sistent low-bob posture along the hopper wall.The significance of this sequence is not yetclear .

Considering all the evidence, a likely con-clusion is that wall pecking has been over-emphasized in superstition paradigms (Fen-ner, 1980; Innis, Simmelhag-Grant, andStaddon, 1983). A careful review of the lit-erature shows that pecking is a frequent pat-tern for no more than half of naive birds ofthe more peck-orientedstrains, and apparent-ly emerges primarily at relatively short in-terreward intervals and after a large numberof sessions. Innis, Simmelhag-Grant, andStaddon (1983) reported little pecking in Car-neaux pigeons at any interval above 6 s, andthis even though they apparently counted someexamples of head bobbing as pecking. It couldbe argued from these data that wall peckingis just. one potential component of wall-di-rected behavior, and frequently a minor one.

Finally, there is some question whether thepecking we did obtain was attributable tostimulus substitution. In many ways it ap-peared more related to search than to inges-tion. In the present experiments, pecking con-sisted of a mixture of downward angled,glancing contacts along the lower part of thewall, and occasional strong sideways swipingof the beak along the floor. The latter move-ment typically occurs in digging and re-arranging the substrate when foraging (Lucas,1981; Schorger, 1955; Whitman, 1919). Skin-ner (1948; see our Table 1) also reported these"brush-type" pecking movements. The wallpecking in films by Reberg and Innis is moreof the driving open-beak variety that has beenobserved directed toward lighted keys (e.g.,Jenkins & Moore, 1973). Even this peckingcan be considered more manipulative thanconsummatory (cf. Delius, 1983), inasmuchas a peck of this force would be more likelyto scatter food than pick it up U. D. Deich &H. P. Zeigler, personal communication, April,1984; see also LaMon, 1981; Zweers, 1982).In sum, although the control of pecking in thesuperstition paradigm remains of interest, wefeel that it is a more complex and variable

behavior than typically has been acknowl-edged.

Interim BehaviorThe present data did support the temporal

organization of behavior reported by Staddonand Simmelhag (1971). However, their cate-gorizationinto exclusive interim and terminalpatterns (or interim, terminal, and facultativepatterns; see Staddon, 1977) appears prob-lematic. For example, our dominant categoryof wall-directed behavior was classified as in-terim behavior in Staddon and Simmelhag's(1971) study. Further, the circling response,classified as interim behavior, sometimes didnot occur (the bird immediatley began wall-directed behavior), or occupied the entire in-terfood interval (when the hopper was in thefloor). Thus, the behavioral organization un-derlying superstitious behavior appears to bemore complex and flexible than the interim-terminal categorization suggests. Similarpoints have been made by Reberg et al. (1978),Innis, Reberg, Mann, Jacobson, and Turton(1983), and Innis, Simmelhag-Grant, andStaddon (1983).

An additional problem with the analysis ofStaddon and Simmelhag (1971) is their as-sumption that interim behavior patterns ar:-eunrelated to the reward. In the present re-suits, the only consistent interim pattern wascircling away from the hopper and returning.Observation of pigeons in small flocks and in-dividual feeding situations reveals that a pi-geon frequently feeds for short periods at aparticular location and then moves away fromthat point continuing to search for other fooditems (personal observation; Murton, 1965).Such immediate postfood searching would beexpected from a seed eater like the pigeon thatfeeds in patches of closely distributed fooditems (Murton, 1965).

Circling appears to be an adjustment of thespecies-typical post food search pattern to lab-oratory situations in which food periodicallyreappears at the same point. When'the hopperwas in the center of the floor approximatinga ground-foraging situation, circling was adominant pattern for most birds. Thus, theinterim pattern of circling away after feedingfrom the hopper appears to be directly relatedto the spatial and temporal organization of thepigeon's feeding rather than to motivationally


complete repertoire of food-related behaviorpatterns that range from searching, throughcapture and preparatory handling, to inges-tion (see Timberlake, 1983b, 1983c). For ex-ample, as the probability of food increases inthe presence of appropriate predictive andsupporting stimuli, we would expect to be ableto condition pigeons to: (1) land near groupsof pigeons in foraging postures (e.g., with ex-posed neck feathers (Murton, 1965 ]); (2) ap-proach areas where pecking sounds and move-ments are displayed; (3) walk about a foodarea in patterns and postures that ordinarilywould facilitate locating further food items;(4) search visually for predictive features inthe substrate; (5) brush and poke at groundcover, thus exposing new search areas (Whit-man, 1919); (6) grasp certain-sized items inthe beak; (7) mandibulate these items in pre-dictable ways that test their suitablity as foodbefore swallowing (Zweers, 1982); and (8) re-ject or accept objects with particular texturesand tastes.

The emergence of each of these patternsshould depend both upon the physical char-acteristic of the stimuli available and the tem-poral relation of these stimuli to food. Thecloser in time to food presentation a particularstimulus occurs, the more likely that the stim-ulus will elicit preconsummatory or ingestiveforms of appetitive behavior based more on itsclose temporal relation to food and less on itsunique physical characteristics. Such resultswould be compatible with, although they donot require, a stimulus-substitution account(Timberlake, 1983c). On the other hand, thefurther away from food in time a predictivestimulus occurs (up to a point), the more like-ly it will be to elicit patterns of search andapproach behavior that are highly dependenton the physical characteristics of the stimulus.Such results are not compatible wtih stimu-lus-substitution accounts (see also Konorski,1967).

It may be worth noting that within thepresent behavior-system view there is no sharpdistinction between appetitive and consum-matory behavior. All complex behavior pat-terns are appetitive in that components varywith stimulus conditions (see Zweers, 1982,for a particularly pretty analysis of the ap-petitive nature of "consummatory" pecking inpigeons). On the other hand, nearly all be-

irrelevant behavior. Further support for thisview is provided by evidence that pigeons re-ceiving water presentations do not circle awayfollowing drinking, but remain near the watersource (Reberg et al., 1978; Innis, Reberg,Mann, jacobson, & Turton, 1983; Innis,Simmelhag-Grant, & Staddon, 1983). Suchbehavior would be expected in the case of acommodity that is found most often at concen-trated spaced locations (e.g., sources of water)rather than distributed in small packages (e.g.,seeds).

A Behavior-System Approach

A more parsimonious account of supersti-tious behavior may be provided within theframework of the appetitive structure of be-havior systems (Timberlake, 1983a, 1983b,1983c, 1984; see also Davey & Cleland, 1984;Davey, Cleland, & Oakley, 1982; jenkins,Barrera, Ireland, & Woodside, 1978; Stad-don, 1983). This approach suggests that theorganization of behavior depends on function-al systems related to important ecological vari-ables. These "behavior systems" can be viewedas collections of organized modules (units ofstimulus sensitivities and related motor pat-terns) that can be collectively primed by con-tact with a reward. When such a behaviorsystem is activated, these modules provide asubstrate of preorganized behavioral compo-nents and stimulus sensitivities from which anecologically adaptive sequence of behavior islikely to emerge. The selection and integrationof modules into an organized pattern of be-havior appear to depend on a number of fac-tors, including elicitation by the physical char-acteristics of predictive and supporting stimuliin the environment, the repetition of a re-sponse sequence in particular stimulus con-ditions, and the instigating role of one modulewith respect to another, as well as the spatialand temporal contingencies imposed by theenvironment (or the experimenter).

The traditional stimulus-substitution ac-count of Pavlovian conditioning (Pavlov, 1927)has assumed that predictive stimuli elicit thesame behavior elicited by the rewarding stim-ulus (e.g., consummatory pecking in the caseof pigeons and food). In contrast to this focuson consummatory behavior, the appetitive-structure view of elicitation assumes the pos-sibility of conditioning a more complex and


the relation of postfood behavior to locatingadditional sources of food.

A second critical question concerns thefunctional basis for wall-directed behavior inthe present experiments. Although we cannotanswer definitely, we note that wall-directedbehavior involves components of food beggingby young squab (cf. Delius,1983), includingstepping, bumping, directed head bobbing, andslightly raised wings. A few such patterns alsomay be seen in female pigeons soliciting court-ship feeding from males. Whether these com-ponents can account completely for wall-di-rected behavior must await the outcome offurther research.

In summary, our results support a view ofsuperstitious behavior as elicited in a broadsense. It seems likely to us that superstitiousbehavior is based on species-typical compo-nents of food-getting behavior, although theirexact functional basis can be inferred readilyonly for circling and pecking. Such an as-sumption may be difficult to prove, but weconsider it a more productive path of investi-gation than the assumption that such behaviorpatterns are accidentally reinforced, or elicitedas conditioned consummatory behavior .

havior patterns have fixed aspects. Whatshould be of interest to learning researchers ishow the variable and fixed aspects of behaviorinteract in producing integrated performance(e.g., Hailman, 1967).

From an ecological viewpoint it is very im-portant that an animal be able to produceadaptive organized sequences of behavior inthe presence of incomplete, novel, and inter-mittent cues. The availability of systems ofpreorganized stimulus sensitivities and re-sponse tendencies should facilitate adaptivebehavior in these environments. In the super-stition paradigm there are no specific predic-tive stimuli, but the pigeon still shows a se-quence of searching and preconsummatorybehavior loosely organized around the presen-tation of food. Locomotor searching is shownearly in the interfood interval, especially whenthe interval is long (Innis, Simmelhag-Grant,& Staddon, 1983) or when omission of foodoccurs within a periodic sequence. More vig-orous wall-directed behavior, including peck-ing when conditions are suitable, occurs morefrequently toward the end of the interfood in-terval when the overall probability of food ishigh (see also Staddon, 1977; Staddon & Sim-melhag, 1971).

Presumably, modules and behavior systemsevolved that produce learning of adaptive be-havior within particular ranges of ecologicallyrelevant conditions Oohnston, 1982). If anunusual or incomplete behavioral context ispresented (as may easily occur in the labora-tory and may also happen in the field), thestructure of appetitive patterns may promotethe learning of behavior that appears mal-adaptive, nonoptimal, or "superstitious." Forexample, circling away from the feeder in thepresent studies was not adaptive because foodwas not available elsewhere.

A number of questions remain unansweredwithin the present approach. For example,over what interfood intervals and durations offeeding can one expect interim behavior to berelated to obtaining food ? Innis, Simmelhag-Grant, and Staddon (1983) reported similarinterim behavior in interfood intervals of upto 300 s. Roper (1980) has shown similar in-terreward behavior under both experimenter-terminated and subject-terminated bouts ofeating and drinking. It seems reasonable toassume that the extent of food-related interimbehavior should depend upon the range ofmeal lengths available to the species, and upon~

REFERENCES

Baerends, G. P. (1976). On drive, conflict, and instinct,and the functional organization of behavior. In M. A.Corner & D. F. Swaitb (Eds.), Perspectives in brainresearch (Vol. 45, pp. 427-447). Amsterdam: Elsev-ier /North-Holland Biomedical Press.

Blaine, C., Innis, N. K., & Staddon, J. E. R. (1980).Stimulus control of behavior induced by aperiodicschedule of food presentation in pigeons. Bulletin of thePsychonomic Society, 16, 131-134.

Davey, G. C. L., & Cleland, G. G. (1984). Food antici-pation and lever-directed activities in rats. Learningand Motivation, 15, 12-36.

Davey, G. C. L., Cleland, G. G., & Oakley, D. A. (1982).Applying Konorski's model of classical conditioning tosignal-centered behavior in the rat: Some functionalsimilarities between hunger CRs and sign-tracking.Animal Learning 6- Behavior, 10, 257-262.

Delius, J. D. (1983). Learning. In M. Abs (Ed.), Phys-iology and behaviour of the pigeon (pp. 327-355). Lon-don: Academic Press.

Fenner, D. (1980). The role of contingencies and "prin-ciples of behavioral variation" in pigeons' pecking.Journal of the Experimental Analysis of Behavior, 34,1-12.

Hailman, J. P. (1967). The ontogeny of an instinct.Behaviour (Suppl. No.15).

Herrnstein, R. J. (1966). Superstition: A corollary ofthe principles of operant conditioning. In W. K. Honig(Ed.), Operant behavior: Areas of research and applica-tion (pp. 33-51). New York: Appleton-Century-Crofts.


Roper, T. J. (1980). Behaviour of rats during self-initi-ated pauses in feeding and drinking, and during per-iodic response-independent delivery of food and water.Quarterly Journal of Experimental Psychology, 32, 459-472.

Schorger,A.W. (1955). The passenger pigeon: Its naturalhistory and extinction. Madison: University of Wiscon-sin Press.

Scott, J. P. (1958). Animal behavior. Chicago: Universityof Chicago Press.

Sheffield, F. D. (1965). Relation between classical con-ditioning and instrumental learning. In W. F. Prokasy(Ed.), Classical conditioning (pp. 302-322). New York:A ppleton -Cen tury -C rof ts .

Skinner, B. F. (1948). "Superstition" in the pigeon.Journal of Experimental Psychology, 38, 168-172.

Staddon, J. E. R. (1977). Schedule-induced behavior.In W. K. Honig & J. E. R. Staddon (Eds.), Handbookof operant behavior (pp. 125-152). Englewood Cliffs,NJ: Prentice-Hall.

Staddon, J. E. R. (1983). Adaptive behavior and learning.Cambridge: Cambridge University Press.

Staddon, J. E. R., & Simmelhag, V. L. (1971). The"superstition" experiment: A reexamination of its im-plications for the principles of adaptive behavior. Psy-chological Review, 78, 3-43.

Timberlake, W. (1983a). Appetitive structure andstraight alley running. In R. L. Mellgren (Ed.), Ani-mal cognition and behavior (pp. 165-222). Amsterdam:North Holland Press.

Timberlake, W. (1983b). The functional organization ofappetitive behavior: Behavior systems and learning. InM. D. Zeiler & P. Harzem (Eds.), Advances in analysisof behaviour: Vol. 3. Biological factors in learning (pp.177-221). Chichester, England: Wiley.

Timberlake, W. (1983c). Rats' responses to amovingobject related to food or water: A behavior-systemsanalysis. Animal Learning & Behavior, 11, 309-320.

Timberlake, W. (1984). An ecological approach tolearning. Learning and Motivation, 15, 321-333.

Timberlake, W., Wahl, G., & King, D. (1982). Stim-ulus and response contingencies in the misbehavior ofrats. Journal of Experimental Psychology: Animal Be-havior Processes, 8, 62-85.

Tinbergen, N. (1951). The study of instinct. Oxford:Clarendon Press.

Whitman, C. 0. (1919); The behavior of pigeons. In H.A. Carr (Ed.), Posthumous Works of Charles Otis Whit-man (Carnegie Institution of Washington, PublicationNo.257 , Vol. 3). Philadelphia: Lippincott.

Williams, D. R. (1981). Biconditional behavior: Con-ditioning without constraint. In C. M. Locurto, H. S.Terrace, & J. Gibbon (Eds.), Autoshaping and condi-tioning theory (pp. 55-99). New York: Academic Press.

Williams, D. R., & Williams H. (1969). Auto-main-tenance in the pigeon: Sustained pecking. despite con-tingent non-reinforcement. Journal of the ExperimentalAnalysis of Behavior, 12, 511-520.

Zweers, G. A. (1982). Pecking of the pigeon (Colum-ba livia L.). Behaviour, 81, 173-230.

Hogan, j. A., & Roper, T. j. (1978). A comparison ofthe properties of different reinforcers. In j. S. Rosen-blatt, R. A. Hinde, C. Beer, & M.-C. Busnel (Eds.),Advances in the study of behavior (Vol. 8, pp. 155-255).New York: Academic Press.

Hollis, K. L. (1982). Pavlovian conditioning of signal-centered action patterns and autonomic behavior: Abiological analysis of function. In j. S. Rosenblatt, R.A. Hinde, C. Beer, & M.-C. Busnel (Eds.), Advancesin the study of behavior (Vol. 12, pp. 1-64). New York:Academic Press.

Innis, N. K., Reberg, D., Mann, B., jacobson, j., &Turton, D. (1983). Schedule-induced behavior forfood and water: Effects of interval duration. BehaviourAnalysis Letters, 3, 191-200.

Innis, N. K., Simmelhag-Grant, V. L., & Staddon, j. E.R. (1983). -Behavior induced by periodic food deliv-ery: The effects of interfood interval. Journal of theExperimental Analysis of Behavior, 39, 309-322.

jenkins, H. M. (1977). Sensitivity of different responsesystems to stimulus-reinforcer and response-reinforcerrelations. In H. Davis & H. M. B. Hurwitz (Eds.),Operant-Pavlovian interactions (pp. 47-66). Hillsdale,Nj: Erlb.1:um.

jenkins, H. M., Barrera, F. j., Ireland, C., & Woodside,B. (1978). Signal-centered action patterns of dogs inappetitive classical conditioning. Learning and Moti-vation, 9, 272-296.

jenkins, H. M., & Moore, B. R. (1973). The form ofthe auto-shaped response with food or water reinfor-cers. Journal of the Experimental Analysis of Behavior ,20,163-181.

johnston, T. D. (1982). Selective costs and benefits inthe evolution of learning. In j. S. Rosenblatt, R. A.Hinde, C. Beer, & M.-C. Busnel (Eds.), Advances inthe study of behavior (Vol. 12, pp. 65-106). New York:Academic Press.

Konorski, j. (1967). ]ntegrative activity of the brain. Chi-cago: University of Chicago Press.

LaMon, B. C. (1981). Conditioned and consummatorygrasping in the pigeon. Unpublished doctoral disserta-tion, Hunter College, City University of New York.

Lucas, G. A. (1981). Some effects of reinforcer avail-ability on the pigeon's responding in 24-hour sessions.Animal Learning & Behavior, 9, 411-424.

Morse, W. H., & Skinner, B. F. (1958). Some factorsinvolved in the stimulus control of operant behavior.Journal of the Experimental Analysis of Behavior, 1,103-107.

Murray, C. S. (1973). Conditioning in Betta splendens.Unpublished doctoral dissertation, University ofPennsylvania.

Murton, R. K. (1965). The wood-pigeon. London: Col-lins.

Pavlov, I. P. (1927). Conditioned reflexes. (G. V. Anrep,Trans.). London: Oxford University Press.

Reberg, D., Innis, N. K., Mann, B., & Eizenga, C.(1978). "Superstitious" behavior resulting from per-iodic response-independent presentations of food orwater. Animal Behaviour, 26,507-519.

Reberg, D., Mann, B., & Innis, N. K. (1977). Super-stitious behavior for food and water in the rat. Physi-ology & Behavior, i9, 803-806.

Received September 21, 1984Final acceptance July 22, 1985

THE BASIS OF SUPERSTITIOUS BEHAVIOR: CHANCE …bsl/Papers from lab/Timberla_1985.pdf · the basis...

Documents

Transcript of THE BASIS OF SUPERSTITIOUS BEHAVIOR: CHANCE …bsl/Papers from lab/Timberla_1985.pdf · the basis...