JOURNAL OF THE EXPERIMENTAL ANALYSIS OF BEHAVIOR

EFFECTS OF DISCRETE-TRIAL AND FREE-OPERANTPROCEDURES ON THE ACQUISITION AND

MAINTENANCE OF SUCCESSIVEDISCRIMINATION IN RATSSHIN HACHIYA AND MASATO ITO

OSAKA CITY UNIVERSITY

Rats were trained on a successive discrete-trial discrimination between two tonal stimuli to examinethe effects of availability of a lever during intertrial intervals. In the discrete-trial condition, in whicha lever was removed from the chamber during intertrial intervals, 10-s trials were initiated by thepresentation of both discriminative stimulus and lever. In the free-operant condition, in which a leverwas present during both trials and intertrial intervals, 10-s trials were initiated only by the presentationof a discriminative stimulus. Experiment 1 employed 50-s intertrial intervals and demonstrated thatdiscriminative performances were acquired faster and maintained better in the free-operant conditionsthan in the discrete-trial conditions. Experiment 2 employed 5-s intertrial intervals and showed thatpoor discriminative performances in the discrete-trial conditions were improved. These results indicatethat the presentation of a lever to start a trial can overshadow or mask the control by a discriminativestimulus and thereby obstruct the acquisition and maintenance of discriminative performances. Fur-thermore, the overshadowing or masking effects are strengthened as a function of the duration ofintertrial intervals.

Key zwords: successive discrimination, discrete-trial procedure, free-operant procedure, overshad-owing, masking, lever press, rats

Two different procedures, discrete-trial andfree-operant procedures, have been employedtraditionally in the study of learning. The dis-crete-trial procedure is a procedure in whichdiscrete occurrences of a specific, externallycontrolled stimulus event (e.g., the insertion ofa lever in an operant chamber or the openingof a door in a runway) and/or a discriminativestimulus enable an organism to make a des-ignated response. The free-operant procedureis a procedure in which the opportunity tomake a designated response is freely availableto the organism.

Logan and Ferraro (1970) presented aframework for conceptualizing the relation-ships between the discrete-trial and free-op-erant procedures. In their framework, a dis-criminative stimulus signaling the consequenceof responding is distinguished from an en-abling stimulus, which refers to an external

Shin Hachiya is now at the Faculty of Commerce, Uni-versity of Marketing and Distribution Sciences. Corre-spondence and reprint requests may be sent to Shin Ha-chiya, Faculty of Commerce, University of Marketing andDistribution Sciences, 3-1, Gakuen-Nishimachi, Nishi-Ku, Kobe 673, Japan, or Masato Ito, Department of Psy-chology, Osaka City University, 3-3-138, Sugimoto, Su-miyoshi-Ku, Osaka 558, Japan.

stimulus event that physically permits a subjectto emit the response in question. Within thisframework, the typical discrete-trial procedurediffers from the typical free-operant procedurein the way of presenting the enabling stimulus.For example, in the typical discrete-trial pro-cedure using a runway, the opening of a door(i.e., an enabling stimulus) occurs only at thebeginning of a trial, whereas in the typicalfree-operant procedure using a chamber, anenabling stimulus such as a lever or a key ispresented throughout a session. In other words,the two procedures differ in the availability ofan enabling stimulus during trials and inter-trial intervals (ITIs): Relatively long ITIs(during which responses are not permitted) arealternated with trials (during which only oneresponse is permitted) in the typical discrete-trial procedure, whereas responses are emittedfreely throughout a session in the typical free-operant procedure. Thus, the two proceduresare located at opposite ends of the continuumof the availability of the enabling stimulus.

Although Logan and Ferraro (1970) de-scribed the runway situation as the typical dis-crete-trial procedure, it is easy to arrange adiscrete-trial procedure using lever pressing orkey pecking in an operant chamber (e.g., At-

3

1991, 55, 3-10 NUMBER 1 (JANUARY)

SHIN HACHIYA and MASATO ITO

nip, 1977; Brown & Jenkins, 1968; Platt,1971). In these studies, the presentation of anenabling stimulus and/or a discriminativestimulus defines a trial: A response lever, whichis retracted from a chamber during ITIs, ispresented during trials, or a response key in apigeon chamber, which is darkened during ITIs(i.e., blackout), is lighted during trials, whereasonly a discriminative stimulus is presentedduring trials when a response lever or key isaccessible during trials and ITIs. According toLogan and Ferraro's framework, these pro-cedures are also situated at different positionsin the continuum of the availability of the en-abling stimulus. Therefore, these proceduresmay have different behavioral effects.

There are few experiments that have ex-amined the effect of availability of an enablingstimulus on the acquisition and maintenanceof learned behavior (e.g., Asano, 1976; Taus& Hearst, 1970). Taus and Hearst varied theduration of a blackout interval between pre-sentations of a 30-s stimulus in the presencFof which responses were reinforced on a vari-able-interval schedule of reinforcement, andfound that pigeons' response rates increased inthe presence of the stimulus as a direct functionof the duration of the intervening blackout.This result suggests that response rate is en-hanced as the period in which an enablingstimulus is not available is increased.

Moreover, Asano (1976), with Japanesemonkeys, examined the effects of the avail-ability of an enabling stimulus during ITIs onthe formation of a successive discrimination.Monkeys were first trained to discriminate be-tween red and white discs in a discrete-trialprocedure in which the lever was availableonly during trials. Then the procedure waschanged to a free-operant procedure in whichthe lever was available during both trials andITIs. Discriminative performance did not de-velop under the discrete-trial procedure butdid under the free-operant procedure.The present experiments, using rats as sub-

jects, further examined the effects of the dis-crete presentation of an enabling stimulus onthe acquisition and maintenance of a successivediscrimination. Experiment 1 compared a free-operant condition with a discrete-trial condi-tion: In the free-operant condition a lever waspresent in a chamber during both 10-s trialsand 50-s ITIs, whereas in the discrete-trialcondition a lever was removed from the cham-

ber during 50-s ITIs. Experiment 2 examinedthe effects of the two conditions with ITIsshortened to 5 s.

EXPERIMENT 1METHOD

SubjectsTen experimentally naive male albino rats

of the Wistar strain, approximately 3 monthsold at the beginning of the experiment, weremaintained at about 80% of their free-feedingbody weights throughout the experiment. Theywere housed in individual home cages withcontinuous access to water.

ApparatusThe conditioning chamber was 25 cm by 25

cm by 31 cm. The ceiling and two sidewallswere made of transparent Plexiglas, and theremaining two walls were of sheet metal. Thelever (2.5 cm wide, 0.2 cm thick) was located6.0 cm from the right sidewall and 5.0 cm fromthe grid floor, and extended 2.0 cm from thefront wall. The lever could be retracted by a24-V DC solenoid. A minimum force of about0.15 N was required to operate the lever. Afood pellet (about 45 mg) served as a reinforcerand was delivered into a recess (5.0 cm by 5.0cm by 2.5 cm) located 1 cm above the grid flooron the front wall. A speaker mounted behindthe front wall was used to deliver a 500-Hztone as a positive stimulus (S+) and a 2,000-Hz tone as a negative stimulus (S-) at anintensity of about 85 dB. The illumination inthe chamber was provided by one 24-V DCceiling lamp. The chamber was enclosed in asound-attenuating chest, and masking noisewas provided by an exhaust fan. A logic mod-ule system, located in an adjacent room, con-trolled and recorded the experimental events.

ProcedureA schematic representation of the discrete-

trial (DT) and free-operant (FO) conditionsis shown in Figure 1. Both conditions con-tained trials and ITIs. The maximum durationof a trial was 10 s, and the maximum durationof an ITI was 60 s. Each trial was initiatedby the presentation of a discriminative stim-ulus (i.e., 500-Hz tone, S+; 2000-Hz tone,S-). In the DT condition, the lever that hadnot been presented during the ITI was pre-

DISCRETE-TRIAL AND FREE-OPERANT PROCEDURES

sented with a discriminative stimulus at thebeginning of a trial. A response during an S+trial produced the immediate delivery of a pel-let, retraction of the lever, and termination ofthe discriminative stimulus. A response duringan S- trial terminated the discriminative stim-ulus and retracted the lever without the pre-sentation of a pellet. If a response did not occurduring a trial, the discriminative stimulus andlever were withdrawn at the end of the trial.The FO condition was the same as the DTcondition except that a lever was presentedduring both trials and ITIs in the FO condi-tion. In the FO condition, each trial was ini-tiated by the presentation of a discriminativestimulus, and a response during a trial ter-minated the discriminative stimulus with orwithout the presentation of a pellet on S+ andS- trials, respectively.

Preliminary training. Subjects were initiallytrained to press the lever by the method ofsuccessive approximations. During the nextfour sessions, every response was reinforceduntil 80 reinforcers had been collected. Thenthe subjects were divided into two groups of 5subjects each, called the DFD (i.e., DT-FO-DT) and FDF (i.e., FO-DT-FO) groups; theonly difference was in the order of exposureto the conditions. Each group was trained topress a lever in the presence of an S+ stimulus.This S+-only training was conducted underthe DT condition for the DFD group and theFO condition for the FDF group. For bothgroups, training consisted of 60 trials and con-tinued for six sessions.

Discrimination training. After preliminarytraining, both groups were given 45 sessionsof discrimination training in an ABA design.For the DFD group, discrimination trainingwas first conducted under the DT conditionfor 20 sessions, next under the FO conditionfor 15 sessions, and finally under the DT con-dition again for 10 sessions. The order of theDT and FO conditions was reversed for theFDF group. Each discrimination session con-sisted of 30 S+ and 30 S- trials, alternatedaccording to a Gellerman sequence. A discrim-ination ratio was calculated by dividing thenumber of responses during S+ trials by thetotal responses during both S+ and S- trials.

RESULTS AND DISCUSSIONTable 1 shows the mean number of re-

sponses per session on S+ and S- trials in

STIMULUS

LEVER

RESPONSEnro rr

STIMULUS

DT50 + t SEC sL.-

s+ FS-FOIt, 50+ t SEC .

I

LEVER J

RESPONSE 11 1 1 1 1 1 1

PELLET

Fig. 1. Diagram of relations between discriminativestimuli, lever availability, responding, and reinforcementunder the DT and FO conditions. Upward displacementof lines indicates the onset of event.

the DT and FO conditions for the DFD andFDF groups. The data were averaged over thelast five sessions in each condition. All ratsemitted a response on almost all S+ trials ofboth DT and FO conditions, whereas theyresponded less often on S- trials of the FOcondition than on those of the DT condition.

Figure 2 shows the discrimination ratios andthe number of intertrial responses under theDT and FO conditions for both groups. Forthe DFD group in the first DT condition, cleardifferential responding was observed for only1 rat (R428); discrimination ratios ranged from.58 to .77 in the last session. However, whenshifted to the FO condition, all discriminationratios for all rats abruptly increased and wereat least .75 in the last session. In the secondDT condition, discrimination ratios were at-tenuated temporarily and recovered slowly inthe later sessions.

For the FDF group, the discrimination ra-tios showed clear differential responding bythe end of the first FO condition. The discrim-ination ratios were attenuated after the tran-sition to the DT condition and again recoveredin the second FO condition. Furthermore, theresults of the first condition in both groupsshowed that differential responding developedfaster under the FO condition than under theDT condition. All rats of both groups alsoemitted intertrial responses in the FO condi-tions. There were, however, no systematicchanges in the number of intertrial responses.

This experiment thus demonstrated that thediscrete presentation of an enabling stimulus(i.e., a lever) has a deleterious effect on theacquisition and maintenance of a successive

5

PELLETI

SHIN HACHIYA and MASATO ITO

Table 1The mean number of responses per session on S+ and S- trials in the discrete-trial (DT) andfree-operant (FO) conditions for the DFD and FDF groups in Experiment 1. The data wereaveraged over the last five sessions in each condition. The standard deviation is presented inparentheses.

DFD goup DT FO DTSubject S+ S- S+ S- S+ S-

R424 29.0 (1.6) 20.0 (2.0) 29.6 (0.5) 0.6 (0.8) 28.6 (2.3) 12.4 (5.4)R425 29.8 (0.4) 19.6 (4.4) 30.0 (0.0) 3.2 (2.0) 29.8 (0.4) 10.6 (2.3)R426 30.0 (0.0) 17.8 (4.6) 30.0 (0.0) 9.0 (3.9) 30.0 (0.0) 15.2 (3.5)R427 29.8 (0.4) 21.2 (0.8) 30.0 (0.0) 2.4 (1.7) 30.0 (0.0) 21.4 (3.2)R428 30.0 (0.0) 7.6 (1.5) 30.0 (0.0) 2.4 (1.4) 30.0 (0.0) 8.6 (3.3)M 29.7 (0.4) 17.2 (4.9) 29.9 (0.2) 3.5 (2.9) 29.7 (0.5) 13.6 (4.4)

FDF group FO DT FOSubject S+ S- S+ S- S+ S-

R419 29.6 (0.5) 1.2 (0.4) 29.8 (0.4) 21.2 (3.5) 29.8 (0.4) 0.8 (0.8)R421 28.4 (1.2) 5.2 (3.0) 30.0 (0.0) 30.0 (0.0) 29.8 (0.4) 6.4 (2.1)R10 29.8 (0.4) 0.6 (0.8) 30.0 (0.0) 27.4 (0.5) 30.0 (0.0) 0.8 (0.8)R12 29.8 (0.4) 0.6 (0.5) 29.8 (0.4) 5.4 (2.1) 30.0 (0.0) 0.2 (0.4)R422 28.6 (2.8) 1.6 (1.0) 29.8 (0.4) 4.0 (1.4) 29.8 (0.4) 1.0 (0.6)M 29.2 (0.6) 1.8 (1.7) 29.9 (0.1) 17.6 (10.9) 29.9 (0.1) 1.8 (2.3)

discrimination. These results are consistentwith those obtained in Asano's (1976) studyin spite of some procedural differences; in Asa-no's study monkeys discriminated between twovisual stimuli with ITIs of 30 s, whereas inthe present experiment rats discriminated be-tween two tonal stimuli with ITIs of 50 s. Thepresent results thus confirmed the generalityof Asano's results.

EXPERIMENT 2The discrete-trial procedure should come to

approximate the free-operant procedure whenthe ITI (during which an enabling stimulusis not presented) is reduced to zero. If so, thedeleterious effects of the discrete presentationof a lever should be attenuated with the ITIsshortened. To examine this possibility, Ex-periment 2 was conducted with ITIs of 5 s.

METHODSubjects and Apparatus

Five experimentally naive male albino ratsof the Wistar strain, approximately 3 monthsold at the beginning of the experiment, weremaintained at about 80% of their free-feedingbody weights. The same apparatus was usedas in Experiment 1.

ProcedureThe procedure was the same as that used

in Experiment 1 except for the use of 5-s ITIs.The subjects were divided into the DFD andFDF groups and received discriminationtraining for 45 sessions in an ABA design.Discriminative performance was evaluated bythe discrimination ratio obtained from dividingthe number of responses during S+ trials bythe total responses during S+ and S- trials,as in Experiment 1.

RESULTS AND DISCUSSIONTable 2 shows the mean number of re-

sponses per session on S+ and S- trials inthe DT and FO conditions for the DFD andFDF groups. The data were averaged over thelast five sessions in each condition. On almostall S+ trials of the DT and FO conditions, all5 rats emitted a response. However, all but 1rat (R1) responded less often on S- trials ofthe FO condition than on those of the DTcondition.

Figure 3 shows the discrimination ratios andthe number of intertrial responses under theDT and FO conditions for both groups. Forall 3 rats in the DFD group, discriminationratios increased gradually toward a value of.88 or higher in the last session of the first DTcondition. Two rats (R4 and R1) maintained

DISCRETE-TRIAL AND FREE-OPERANT PROCEDURES

DT1.0

.7

.5

.41.0

.7

~- .5< .4= 1.0

M° .7

.4E1.0

cn .7

.5

.41.0

.7

.5

.4

FO DT FO DT FO

Lc

LUJ

I .

__o

LU

1 10 20 30 40 45 1 10 20 30 40 43

TRAINING SESSIONSFig. 2. Discrimination ratios (calculated by dividing the number of responses during S+ trials by the total responses

during both S+ and S- trials) obtained from each session of the DT and FO conditions and the number of intertrialresponses obtained from each session of the FO conditions. Solid lines indicate discrimination ratios, and broken linesindicate the number of intertrial responses. The data for the DFD and FDF groups are presented in the left and rightpanels, respectively.

Table 2

The mean number of responses per session on S+ and S- trials in the discrete-trial (DT) andfree-operant (FO) conditions for the DFD and FDF groups in Experiment 2. The data wereaveraged over the last five sessions in each condition. The standard deviation is presented inparentheses.

DFD group DT FO DTSubject S+ S- S+ S- S+ S-

R3 30.0 (0.0) 1.4 (0.8) 30.0 (0.0) 0.8 (0.7) 30.0 (0.0) 12.6 (2.2)R4 29.8 (0.4) 2.4 (1.5) 30.0 (0.0) 1.4 (0.8) 30.0 (0.0) 1.8 (1.3)Rl 30.0 (0.0) 0.8 (0.7) 30.0 (0.0) 1.0 (0.9) 30.0 (0.0) 1.2 (1.2)M 29.9 (0.1) 1.5 (0.6) 30.0 (0.0) 1.1 (0.2) 30.0 (0.0) 5.2 (5.2)

FDF group FO DT FOSubject S+ S- S+ S- S+ S-

R2 30.0 (0.0) 2.8 (1.5) 30.0 (0.0) 29.4 (1.2) 30.0 (0.0) 18.2 (4.2)R5 29.8 (0.4) 2.2 (1.2) 30.0 (0.0) 6.4 (2.1) 30.0 (0.0) 4.0 (2.0)M 29.9 (0.1) 2.5 (0.3) 30.0 (0.0) 17.9 (11.5) 30.0 (0.0) 11.1 (7.1)

SHIN HACHIYA and MASATO ITO

DT FO DT

CD

L-J

.u:/

:L

__

1 10 20 30 40 45 1 10 20 30 40 45

TRAINING SESSIONSFig. 3. Discrimination ratios (calculated by dividing the number of responses during S+ trials by the total responses

during both S+ and S- trials) obtained from each session of the DT and FO conditions and the number of intertrialresponses obtained from each session of the FO conditions. Solid lines indicate discrimination ratios, and broken linesindicate the number of intertrial responses. The data for the DFD and FDF groups are presented in the left and rightpanels, respectively.

this high level of discriminative performancethroughout the FO and second DT conditions.The remaining rat's (R3) discrimination ratiosdecreased when the procedure was shifted fromthe FO to the second DT condition.

For the FDF group, the discriminative per-formances were acquired in the first FO con-dition. However, discrimination ratios wereattenuated in the DT condition and then re-covered in the second FO condition. Further-more, the results of the first condition in bothgroups showed that discrimination ratios in-creased a little faster in the FO condition thanin the DT condition; the ratios were at least.68 in the first session of the first FO conditionfor the FDF group, but they were at most .57in the first session of the first DT conditionfor the DFD group. However, the level ofdiscriminative performance was about the samein the last sessions of the two conditions. Al-though the rats were permitted to respond dur-ing 5-s ITIs in the FO condition, all but 1 rat(R2) emitted few intertrial responses.The present experiment with 5-s ITIs thus

demonstrated a smaller difference in discrim-inative performance between the DT and FOconditions than did Experiment 1 with 50-sITIs: When the condition was shifted from the

DT to FO condition and vice versa, only 3 of5 rats (R3, R2, and R5) showed large changesin discrimination ratios in the present ex.per-iment, whereas all 10 rats showed such changesin Experiment 1; the between-group compar-ison of the first DT and FO conditions alsoshowed that the differences between the twoconditions were smaller in the present exper-iment than in Experiment 1 (compare Figures2 and 3). These results depended on the dif-ferences between discrimination ratios in theDT conditions of Experiments 1 and 2, be-cause the discrimination ratios in the FO con-ditions were much the same in both experi-ments. Thus, the deleterious effects of thediscrete presentation of an enabling stimuluswere attenuated when the duration of the ITI(during which the enabling stimulus was notpresented) was reduced.

GENERAL DISCUSSIONLogan and Ferraro (1970) pointed out that

the typical discrete-trial procedure differedfrom the typical free-operant procedure in theavailability of an enabling stimulus. The pres-ent experiments manipulated the availabilityof a lever by removing or presenting it during

DISCRETE-TRIAL AND FREE-OPERANT PROCEDURES 9

the ITI and examined the effects of this avail-ability on the acquisition and maintenance ofa successive discrimination between two tonalstimuli. Experiments 1 and 2 demonstratedthat rats' discriminative performances were ac-quired faster and were maintained better whenthe lever was presented during both the trialsand the ITIs (i.e., the FO condition) than whenthe lever was presented only during the trials(i.e., the DT condition). Moreover, exami-nation of the results revealed that poor dis-criminative performances in the DT conditionwere improved when the duration of ITIs wasreduced from 50 s in Experiment 1 to 5 s inExperiment 2, although discriminative per-formances in the FO conditions were much thesame in the two experiments. Thus, these ex-periments indicate that discriminative perfor-mance varies as a function of availability of anenabling stimulus.The present results can be understood in

terms of an overshadowing effect (in which thepresence of a more intense or salient stimulusinterferes with the acquisition of stimulus con-trol by a less intense or salient stimulus) or interms of a masking effect (in which the pres-ence of the former obscures the expression ofstimulus control by the latter; Mackintosh,1977). The difference in discriminative per-formances between the DT and FO conditionsin Experiments 1 and 2 may be interpreted asthe presence or absence of the occurrence ofovershadowing or masking. In the DT con-ditions of both experiments, the lever was pre-sented simultaneously with the tonal discrim-inative stimulus at the beginning of the trialsand therefore may have functioned as a pre-dictive stimulus, as in the studies of autoshap-ing and negative automaintenance with rats(Atnip, 1977; Davey, Oakley, & Cleland, 1981;Myer & Hull, 1974; Stiers & Silberberg, 1974).However, the lever might not have served asa predictive stimulus in the FO conditions ofboth experiments, because the lever remainedin the chamber at all times in these conditions.Thus, the lever may have overshadowed ormasked the stimulus control by the tonal stim-ulus in the DT conditions of both experiments,but not in the FO conditions. Further, thedifference in discriminative performances be-tween the DT conditions in Experiments 1and 2 may be interpreted as the difference inthe degree of overshadowing or masking by thelever. As the duration of the ITI (during which

the lever was not presented) was increased, thelever may have become a stronger predictivestimulus and overshadowed or masked the ton-al discriminative stimulus more completely.

This interpretation is consistent with thepredictions from scalar expectancy theory(Gibbon & Balsam, 1981) and from delay-reduction theory (Fantino, 1977, 1981). Forexample, scalar expectancy theory applied toautoshaping procedures states that the evoca-tive control of a stimulus increases as a func-tion of the ratio of the interreinforcement (i.e.,cycle) duration (C) to the trial duration (T).Applied to the present experiments, the C/Tratio is calculated with respect to the temporalrelation between the reinforcer and the lever.This ratio was larger in the DT conditionsthan in the corresponding FO conditions, be-cause the T value (i.e., the duration of a leverpresentation) was smaller in the DT condi-tions than in the FO conditions, although thesame C value was used in the two conditions.Accordingly, the lever should have more evoc-ative strength and thus be more likely to ob-struct a tonal discrimination in the DT con-ditions. Also, the C/T ratio can be comparedbetween the two DT conditions in Experi-ments 1 and 2. This C/T ratio was larger inthe DT conditions of Experiment 1 than inthose of Experiment 2, because the C valuewas larger in the former conditions than in thelatter conditions, but T values were the samein both conditions. Therefore, the presentationof the lever should have stronger obstructiveeffects on tonal discrimination in the DT con-ditions of Experiment 1 than in those of Ex-periment 2.

REFERENCESAsano, T. (1976). Some effects of a discrete trial pro-

cedure on differentiation learning by Japanese mon-keys. Primates, 17, 53-62.

Atnip, G. W. (1977). Stimulus- and response-reinforcercontingencies in autoshaping, operant, classical, andomission training procedures in rats. Journal of the Ex-perimental Analysis of Behavior, 28, 59-69.

Brown, P. L., & Jenkins, H. M. (1968). Auto-shapingof the pigeon's key-peck. Journal of the ExperimentalAnalysis of Behavior, 11, 1-8.

Davey, G. C. L., Oakley, D., & Cleland, G. G. (1981).Autoshaping in the rat: Effects of omission on the formof the response. Journal of the Experimental Analysis ofBehavior, 36, 75-91.

Fantino, E. (1977). Conditioned reinforcement: Choiceand information. In W. K. Honig & J. E. R. Staddon

10 SHIN HACHIYA and MASATO ITO

(Eds.), Handbook of operant behavior (pp. 313-339).Englewood Cliffs, NJ: Prentice-Hall.

Fantino, E. (1981). Contiguity, response strength, andthe delay-reduction hypothesis. In P. Harzem & M.D. Zeiler (Eds.), Advances in analysis of behaviour (Vol.2): Predictability, correlation, and contiguity (pp. 169-201). Chichester, England: Wiley.

Gibbon, J., & Balsam, P. (1981). Spreading associationin time. In C. M. Locurto, H. S. Terrace, & J. Gibbon(Eds.), Autoshaping and conditioning theory (pp. 219-253). New York: Academic Press.

Logan, F. A., & Ferraro, D. P. (1970). From free re-sponding to discrete trials. In W. N. Schoenfeld (Ed.),The theory ofreinforcement schedules (pp. 111-138). NewYork: Appleton-Century-Crofts.

Mackintosh, N. J. (1977). Stimulus control: Attentionalfactors. In W. K. Honig & J. E. R. Staddon (Eds.),Handbook of operant behavior (pp. 481-513). Engle-wood Cliffs, NJ: Prentice-Hall.

Myer, J. S., & Hull, J. H. (1974). Autoshaping andinstrumental learning in the rat. Journal ofComparativeand Physiological Psychology, 86, 724-729.

Platt, J. R. (1971). Discrete trials and their relation tofree-behavior situations. In H. H. Kendler & J. T.Spence (Eds.), Essays in neobehaviorism (pp. 137-160).New York: Appleton-Century-Crofts.

Stiers, M., & Silberberg, A. (1974). Lever-contact re-sponses in rats: Automaintenance with and without anegative response-reinforcer dependency. Journal oftheExperimental Analysis of Behavior, 22, 497-506.

Taus, S. E., & Hearst, E. (1970). Effects of intertrial(blackout) duration on response rate to a positive stim-ulus. Psychonomic Science, 19, 265-267.

Received April 15, 1987Final acceptance August 1, 1990