· reinforcer (i.e., self-control), a result typically observed with humans. In other words, when the procedures were more typical of nonhu-man self-control procedures—differential

RISKY CHOICE IN PIGEONS AND HUMANS: A CROSS-SPECIES COMPARISON

CARLA H. LAGORIO AND TIMOTHY D. HACKENBERG

UNIVERSITY OF FLORIDA

Pigeon and human subjects were given repeated choices between variable and adjusting delays to tokenreinforcement that titrated in relation to a subject’s recent choice patterns. Indifference curves weregenerated under two different procedures: immediate exchange, in which a token earned during each trialwas exchanged immediately for access to the terminal reinforcer (food for pigeons, video clips forhumans), and delayed exchange, in which tokens accumulated and were exchanged after 11 trials. Theformer was designed as an analogue of procedures typically used with nonhuman subjects, the latter asan analogue to procedures typically used with human participants. Under both procedure types,different variable-delay schedules were manipulated systematically across conditions in ways that alteredthe reinforcer immediacy of the risky option. Under immediate-exchange conditions, both humans andpigeons consistently preferred the variable delay, and indifference points were generally ordered inrelation to relative reinforcer immediacies. Such risk sensitivity was greatly reduced under delayed-exchange conditions. Choice and trial-initiation response latencies varied directly with indifferencepoints, suggesting that local analyses may provide useful ancillary measures of reinforcer value. On thewhole, the results indicate that modifying procedural features brings choices of pigeons and humansinto better accord, and that human–nonhuman differences on risky choice procedures reported in theliterature may be at least partly a product of procedural differences.

Key words: risky choice, reinforcer delay, adjusting-delay procedures, cross-species comparisons, tokenreinforcement, key peck, keyboard response, pigeons, adult humans

_______________________________________________________________________________

Research in the area of risky choice isbroadly concerned with how the value of areinforcer is affected by the variance associat-ed with that reinforcer. Risky choice is typicallystudied by giving subjects choices betweenfixed and variable outcomes of the samearithmetic mean. Preference for the fixed(certain) outcome has been termed riskaversion, and preference for the variable(uncertain) outcome risk proneness. With choic-es between fixed and variable delays toreinforcement, nonhumans are strongly riskprone, preferring the variable delay (see

Mazur, 2004, for a review). This robustsensitivity to reinforcer delay has been foundacross a broad range of species (insects, fish,birds, and mammals; Kacelnik & Bateson,1996), and reinforcer types (e.g., water, food;Bateson & Kacelnik, 1995, 1997; Case, Nichols,& Fantino, 1995; Cicerone, 1976; Davison,1969, 1972; Fantino, 1967; Frankel & vomSaal, 1976; Gibbon, Church, Fairhurst, &Kacelnik, 1988; Herrnstein, 1964; Hursh &Fantino, 1973; Kendall, 1989; Killeen, 1968;Logan, 1965; Mazur, 1984; Navarick & Fantino,1975; O’Daly, Case, & Fantino, 2006; Orduna& Bouzas, 2004; Rider, 1983; Sherman &Thomas, 1968; Zabludoff, Wecker, & Caraco,1988).

Although scant by comparison, laboratoryresearch with human subjects on similarprocedures is mixed and difficult to interpret.Some results show risk aversion (e.g., Kohn,Kohn, & Staddon, 1992), others indifferencebetween the outcomes (e.g., Weiner, 1966). Arecent exception to these findings was report-ed by Locey, Pietras, and Hackenberg (2009).In their study, human participants were givenrepeated choices between fixed and variablereinforcer delays of the same arithmetic meanwith 30-s video clips serving as reinforcers. Thevideos were preselected by the participants,and played only during reinforcement periods

This research was supported by Grant IBN 0420747from the National Science Foundation. An earlier versionof this paper was submitted by the first author to theGraduate School at the University of Florida in partialfulfillment of the Master of Science Degree. Portions ofthese data were presented at the 2006 meeting of theSoutheastern Association for Behavior Analysis and the2007 meetings of the Society for the Quantitative Analysisof Behavior and the Association for Behavior AnalysisInternational. The authors thank Anthony DeFulio,Rachelle Yankelevitz, Leonardo Andrade and BethanyRaiff for their assistance with data collection, and to MarcBranch, Jesse Dallery, and David Smith for comments onan earlier version.

Correspondence should be sent to Carla H. Lagorio,Department of Psychology, P.O. Box 112250, University ofFlorida, Gainesville, Florida 32611-2250 (e-mail: [email protected]).

doi: 10.1901/jeab.2010.93-27

JOURNAL OF THE EXPERIMENTAL ANALYSIS OF BEHAVIOR 2010, 93, 27–44 NUMBER 1 (JANUARY)

27

(i.e., the tape stopped at all other times). Thevariable-delay distributions were bivalued—one long and one short delay, the arithmeticaverage of which equaled the fixed delay. Thefixed delay was systematically varied acrossconditions: 15 s (1 s and 29 s variable delays),30 s (1 s and 59 s variable delays) and 60 s (1 sand 119 s variable delays). Similar to resultswith nonhumans but dissimilar to previousresults with human participants, 3 of 4 subjectsstrongly preferred the variable delays toreinforcement.

In a second part of the study, the 3participants who preferred the variable rein-forcer delays were given repeated choicesbetween different variable-delay schedulescomposed of different distribution types (ex-ponential, bimodal, rectangular and normal).In general, the choice patterns were orderedwith respect to distribution type, with strongerpreference for the distributions with a higherproportion of short delays. The results wereconsistent with the nonlinear discounting ofdelayed reinforcement seen with other ani-mals (e.g., Mazur, 1986a).

The disproportionate weighting of shortdelays reported by Locey et al. (2009) mayhave depended on the use of video rather thantoken (point or money) reinforcers. This isconsistent with the results of other studies withhuman participants showing enhanced sensi-tivity to reinforcer delay with video reinforcers(Hackenberg & Pietras, 2000; Navarick, 1996).Unlike token reinforcers such as points, whosereinforcing efficacy depends on later ex-change for other reinforcers, video reinforcersare ‘‘consumed’’ as they are earned. This maylend greater time urgency to video reinforcersas compared to the more conventional token-type reinforcer, bringing human behavior intogreater alignment with that seen in otheranimals. If correct, this interpretation impliesthat previous differences in risky choicebetween human and nonhuman subjects maybe more a function of procedural variablesthan actual species differences.

Additional support for this hypothesis wouldcome from studying nonhuman choices insituations more akin to the token-type rein-forcement systems commonly employed inresearch with humans. For example, in a self-control task, Jackson and Hackenberg (1996)gave pigeons repeated choices between asingle token delivered immediately and three

tokens delivered after a brief delay. Tokenscould be exchanged for food during sched-uled exchange periods later each trial. Acrossconditions, the delay to the exchange periodwas manipulated. When the exchange periodwas scheduled immediately following eachtoken delivery, the pigeons preferred thesmaller–sooner reinforcer (i.e., impulsivity), aresult typically observed with nonhumans.When the exchange period was scheduledafter a fixed delay from either choice, however,preference reversed in favor of the larger–laterreinforcer (i.e., self-control), a result typicallyobserved with humans. In other words, whenthe procedures were more typical of nonhu-man self-control procedures—differential de-lays to food—more ‘‘animal-like’’ perfor-mance was seen, but when the procedureswere more typical of human self-controlprocedures—with delays to terminal reinforc-ers held constant—more ‘‘human-like’’ per-formance was seen. This lends support to theidea that procedural variables, such as thedelay to the exchange period, may be respon-sible for cross-species differences in choicebehavior.

The present study sought to bring intogreater alignment the procedures used withhumans and nonhuman subjects in a riskychoice context. It involved explicit compari-sons across (a) species (humans and pigeons),and (b) procedures (consumable-type andtoken-type reinforcers). Choices by pigeons(Experiment 1) and humans (Experiment 2)produced tokens exchangeable for consum-able-type reinforcers: mixed-grain for pigeonsand preferred video clips for humans. Theprocedures were structured in ways thatmimicked consumable-type and token-typereinforcement systems. This was accomplishedby manipulating the exchange delay—the timebetween earning a token and exchanging it forthe terminal reinforcer. In some conditions(immediate exchange), a token could be ex-changed for the reinforcer as soon as it wasearned. In other conditions (delayed exchange),groups of 11 tokens had to be earned prior toexchange for reinforcement. Immediate-ex-change conditions are more akin to consum-able-type reinforcement procedures, permit-ting immediate consumption of the reinforcer.Delayed-exchange conditions are more akin totoken-type reinforcement systems, in whichthe exchange period is scheduled after a

28 CARLA H. LAGORIO and TIMOTHY D. HACKENBERG

group of tokens has been earned. If choicepatterns depend at least in part on theseprocedural variables, then we would expectchoices of both species to be more risk prone(preferring variable outcomes) under imme-diate-exchange than under delayed-exchangeconditions.

Risk sensitivity in both experiments wasassessed with an adjusting-delay procedure, inwhich the delay to a constant (standard)alternative changes systematically in relationto a subject’s recent choice patterns. Subjectschose between a bi-valued (mixed-time) andan adjusting delay to token reinforcement.The main manipulation concerned changes inthe economic context; or more specifically, thedelay between earning a token and exchang-ing it for food (Experiment 1) or video(Experiment 2). Under immediate-exchangeconditions, exchange periods were scheduledjust after each token was earned, permittingimmediate exchange and consumption of thereinforcer. Under delayed-exchange condi-tions, exchange periods were scheduled after11 tokens had been earned, requiring accu-mulation of tokens for later exchange andconsumption. The goal under both proce-dures was to establish indifference points—points of subjective equality—between variableand adjusting reinforcer delays.

A second manipulation concerned thedelays comprising the variable schedule. Thesewere changed in such a way to alter the relativereinforcer immediacies while holding constantthe mean reinforcer delay. Choice patterns ofboth species under both procedure types wereevaluated in relation to Mazur’s (1986b)model:

V ~Xn

i~1

PiA

1 z KDið Þ

� �ð1Þ

where the subjective value (V) of a delayedoutcome that delivers one of n possible delayson a given trial, is a function of the amount ofreinforcement (A), the delay to the outcome(D), and sensitivity to changes in delay (K ). Piis the probability that a delay of Di seconds willoccur. This model predicts that the value orstrength of a reinforcer decreases as the delayto reinforcement increases, with the strengthof a variable delay computed by taking theaverage value weighted by the probability ofoccurrence of all programmed delays.

Equation 1 has provided a good descriptionof prior results on adjusting delay procedures,including the risky-choice variant used here(Mazur, 1986b). The Locey et al. (2009)results, described above, were broadly consis-tent with the delay-based calculation of rein-forcer value proposed by Equation 1, in thatthe model reliably predicted the direction ofpreference for all subjects. More precisepredictions of the model were precluded,however, by the near exclusive preferencesgenerated by the experimental procedures.

Titrating procedures, like those used in thepresent study, have been shown to producedynamic and graded measures of preference,permitting a sharper quantitative assessmentof temporal discounting. In addition, theseprocedures have been used effectively inlaboratory studies of reinforcer discountingin a range of species (Mazur, 1987, 2007;Rodriguez & Logue, 1988), and are thereforewell suited to the present cross-species analysis.Combining these procedures with the manip-ulations of economic context enables a quan-titative comparison of risky choice acrossprocedure type and species.

EXPERIMENT 1METHOD

Subjects

Four experimentally naı̈ve male White Car-neau pigeons (Columba livia), numbered 75,995, 967 and 710, served as subjects. All weremaintained at approximately 83% of their free-feeding weights. Pigeons were housed individ-ually in a colony room, with continuous accessto grit and water outside of the experimentalsessions. The colony room was lighted on a16:8 hr light/dark cycle.

Apparatus

One Lehigh Valley ElectronicsH operantchamber (31 cm long, 35 cm wide, 37 cmhigh), with an altered control panel was used.The panel displayed three horizontally-spacedresponse keys, each 2.5 cm in diameter andmounted 5.5 cm apart, requiring a force ofapproximately 0.26 N to operate. Each side keycould be illuminated to produce green, red, oryellow colors; the center key could be red orwhite. Twelve evenly-spaced red stimuluslamps served as tokens. The tokens werepositioned 2.5 cm above the response keys

CROSS-SPECIES ANALYSIS OF RISKY CHOICE 29

and arranged horizontally at 0.8 cm apart. A0.1-s tone accompanied each token illumina-tion and darkening. A square opening located9 cm below the center response key and 10 cmabove the floor allowed for access to mixedgrain when the food hopper was raised. Thehopper area contained a light that wasilluminated during food deliveries, and aphotocell that allowed for precisely timedaccess to grain. The chamber also containeda houselight, aligned 5 cm above the centerresponse key. A sound-attenuating box en-closed the chamber, with ventilation fans andwhite noise generators continuously runningto mask extraneous noise during sessions.Experimental contingencies were arrangedand data recorded via a computer pro-grammed in MedState Notation Languagewith Med-PCH software.

Procedure

Training. Before exposure to the experimen-tal contingencies, the pigeons received severalweeks of daily sessions on a backward-chainingprocedure designed to train token productionand exchange. Following magazine and key-peck shaping, pecking the center (red) key inthe presence of one illuminated token turnedoff the token light, produced a 0.1-s tone, andraised the hopper for 2.5 s. Once thisexchange response was established, pigeonswere required to peck one or the other sidekey (illuminated either green or yellow withequal probability each cycle) to produce(illuminate) a token. Token production wasaccompanied by a 0.1-s tone and led immedi-ately to the exchange period, during which theside keys were darkened and the center redkey was illuminated. As before, a single peckon the center exchange key turned off thetoken light, produced a 0.1-s tone and raisedthe hopper for 2.5 s. When this chain ofresponses had been established, the nexttraining phase required 11 tokens to accumu-late (by making 11 responses on the illumi-nated token-production keys) before advanc-ing to the exchange period. The side positionof the illuminated token-production key con-tinued to occur randomly after each tokenpresentation. Tokens were always producedfrom left to right and withdrawn from right toleft.

Experimental procedure. Figure 1 shows aprocedural schematic of contingencies on a

typical choice trial, in which subjects chosebetween variable (mixed-time, MT) and ad-justing time delays. The first two trials in asession were forced-choice trials designed toprovide exposure to both choice alternatives.These trials were the same as free-choice trialsexcept that only one response key was illumi-nated and effective. Thereafter, when eitheralternative was selected on two consecutivefree-choice trials, a forced-choice of the other,nonchosen, alternative was presented. Theremaining trials were free-choice trials, inwhich both options were available. These trialsbegan with a trial-initiation response, requir-ing a response on the center white key, afterwhich both side keys were illuminated: onewith green light and the other with yellowlight. The position of the alternatives wasdetermined randomly each trial.

Throughout the experiment, a single re-sponse on the green response key initiated theadjusting delay, and a single response on theyellow response key initiated the MT delay. Forthe duration of the delays, the chosen keycolor remained illuminated while the house-light and other key alternative were darkened.

Fig. 1. Procedural schematic of a free-choice trial inboth the immediate and delayed exchange conditions.Blackened keys indicate that they are not illuminated.Tokens are always illuminated red. Reinforcer is 2.5-saccess to food. W 5 white, Y 5 yellow, G 5 green, R 5 red,MT 5 Mixed Time.


After the delay had elapsed, one tokenstimulus light and the houselight were illumi-nated. During the exchange periods, whereinsubjects could ‘‘trade in’’ earned tokens foraccess to food, both side keys were darkenedand the center red key was illuminated. Oneresponse on the center key turned off a singletoken light and produced the 2.5-s access tomixed grain. When all tokens had beenexchanged, a new trial began.

Each selection of the MT alternative de-creased the adjusting delay on the subsequenttrial by 10%. Conversely, each selection of theadjusting alternative increased by 10% theadjusting delay on the subsequent trial. Theadjusting delay had a lower limit of 1 s and anupper limit of 120 s, and was unaffected byforced-choice responses.

There were two independent variables in theexperiment: (1) MT distribution, and (2)exchange delay. The MT distribution consistedof four values, all with the same overall meaninterfood interval of 30 s: (a) 1 s, 59 s; (b) 10 s,50 s; (c) 20 s, 40 s; and (d) 30 s, 30 s (or FT 30s). The exchange delay consisted of two values:immediate and delayed. In immediate condi-tions, the exchange period commenced im-mediately after each token was earned. Indelayed conditions, the exchange period didnot occur until 11 tokens had been earned(from free- and forced-choice trials com-bined). In these delayed-exchange conditions,all 11 tokens were exchanged for reinforcers insuccession, each requiring one exchangeresponse on the center red key per token.Each session consisted of 44 trials, hence, 44food deliveries. The only difference betweenimmediate and delayed-exchange was in thetiming and number of exchange periods: 44exchange periods with one reinforcer perexchange (immediate exchange) or 4 ex-change periods with 11 reinforcers per ex-change (delayed exchange).

The MT distribution was varied systemati-cally across experimental conditions, firstunder immediate exchange conditions, thenunder delayed exchange conditions. Theselatter conditions also included interspersedreplications of immediate-exchange condi-tions at each MT value to facilitate compari-sons of the immediate and delayed exchange.To provide comparable starting points, eachexperimental condition began with the adjust-ing delay set to 30 s (equivalent to the mean

interreinforcer interval of the MT schedule).Most conditions were conducted without anintertrial interval (ITI) separating successivechoice trials, but one condition per subjectalso included an ITI. During this condition,the ITI was at least 10 s in length, and if theadjusting delay was chosen on the previoustrial and was currently shorter than 30 s, theadjusting delay was subtracted from 30 andadded to the 10 s ITI. This requirement heldapproximately equal the overall rate of rein-forcement from both alternatives. Table 1 liststhe sequence and number of sessions percondition.

Sessions were conducted daily, 7 days perweek. Conditions remained in effect for aminimum of 10 sessions and until stableindifference points were obtained. To assessstability, free choice trials in a session (typicallybetween 30 and 35, but variable given thenature of the forced-choice trials) were dividedin half and median adjusting values werecalculated, resulting in two data points persession. Data were judged visually stable withno trend or bounce, and according to thefollowing criteria: (1) the last six sessions (12data points) contained neither the highest norlowest half-session median of the condition;and (2) the mean of the last three sessionscould not differ from the mean of the previousthree sessions, nor could either groupingdiffer from the overall six-session mean bymore than 15% or 1 s (whichever was larger).

RESULTS

Indifference points between the variableand adjusting delays were determined byaveraging the obtained adjusting delays overthe last 12 data points (six sessions) percondition. Indifference points resulting fromthe immediate exchange conditions wereevaluated in relation to Equation 1. Figure 2displays the obtained indifference points forall 4 subjects under immediate-exchangeconditions plotted as a function of the MTdelay, with the MT schedules arranged alongthe x-axis according to their smallest includeddelay. Data from initial and replication condi-tions were averaged in this plot, providing onedata point per subject per condition. Thecurve displays the predicted delays provided byEquation 1 with A 5 100 (an arbitrary valuebecause amount was not manipulated) andparameter K held constant at 1 (a value which


has provided a good description of priorpigeon data; Mazur, 1984, 1986b, 1987). This,too, proved appropriate for our results, withEquation 1 accounting for 95.0% of thevariance in performance. The equation wasparticularly appropriate when accounting fordata obtained in the MT conditions containingone relatively short delay to reinforcement(i.e., MT 1 s, 59 s and MT 10 s, 50 s).

The pooled data of Figure 2 are consistentwith the data for individual subjects. Theindividual-subject indifference points can beseen in Figure 3, along with those from thedelayed-exchange conditions. The delayed-exchange conditions produced higher indif-ference points at each MT value. Similar toperformance in the immediate-exchange con-ditions, indifference points generally in-creased as a function of the smallest delay inthe MT schedule. At the longer delays,indifference points were generally in excessof the mean IRI of 30 s, reflecting increasedchoices for the adjusting delay.

Including a 10-s ITI had little effect onchoice patterns. Indifference points duringthe initial and replication MT 10 s, 50 sconditions averaged 16.8 s without an ITIand 16.2 s with an ITI.

In addition to the choice data, latencymeasures were also analyzed across exchange-delay and MT-delay conditions. Figure 4 showsmean latencies, averaged across subjects and(adjusting and variable) trial types, underimmediate-exchange conditions. The threebars correspond to latencies to respond onthe exchange key, the choice key, and the trial-initiation key. Exchange latencies were rela-

tively short and undifferentiated across condi-tions, but choice and trial-initiation latenciesvaried systematically as a function of the MTschedule: latencies increased as a directfunction of the shortest delay in the MTschedule. These latencies showed some paral-lel to the overall preferences for the differentMT delays, with shorter latencies occurring inconditions with smaller obtained adjustingdelays.

Figure 5 shows comparable latency dataunder delayed-exchange conditions. Overall,a similar pattern emerged, with shorter choiceand trial-initiation latencies in the MT 1 s, 59 s

Table 1

Sequence and number of sessions per condition (in parentheses) for each pigeon inExperiment 1.

Condition 75 995 967 710

FT 30 s Immediate 1(34) 1(25) 1(22) 1(37)5(19) 5(25) 5(36) 5(11)7(24)

MT 1 s, 59 s Immediate 2(12) 2(10) 2(14) 2(14)9(23) 7(13) 7(16) 7(12)

MT 10 s, 50 s Immediate 4(19) 3(35) 3(29) 4(21)11(21) 9(26) 9(15) 9(36)

MT 20 s, 40 s Immediate 3(11) 4(28) 4(25) 3(23)FT 30 s Delayed 6(10) 6(21) 6(42) 6(28)

8(14)MT 1 s, 59 s Delayed 10(27) 8(23) 8(26) 8(27)MT 10 s, 50 s Delayed 12(24) 10(31) 10(37) 10(18)MT 10 s, 50 s Immediate ITI 13(36) 11(26) 11(16) 11(29)

Fig. 2. Mean indifference points as a function of themixed-time (MT) or fixed-time (FT) delay condition for all4 subjects in Experiment 1. The curve displays predictedindifference points from Equation 1 with K 5 1. See textfor additional details.


condition than in the conditions with longerreinforcer delays. In comparing across Fig-ures 4 and 5, note the differently scaled axes,showing the generally longer latencies under

delayed-exchange than immediate-exchangeconditions.

For delayed-exchange conditions, choiceand trial-initiation latencies were also analyzed

Fig. 3. Mean indifference points for each subject from immediate-exchange (black bars) and delayed-exchange (greybars) conditions as a function of the mixed-time (MT) or fixed-time (FT) delay. Error bars represent standard deviations.

Fig. 4. Mean latencies to respond on exchange,choice, and trial initiation keys as a function of themixed-time (MT) or fixed-time (FT) delay, averaged acrosssubjects, in immediate-exchange conditions. Error barsindicate across-subject standard deviations.

Fig. 5. Mean latencies to respond on exchange,choice, and trial initiation keys as a function of themixed-time (MT) or fixed-time (FT) delay, averaged acrosssubjects, in delayed-exchange conditions. Error barsindicate across-subject standard deviations.


across trials leading to an exchange period.The results are displayed in Figures 6 and 7,respectively, with the dashed lines providing areference point under immediate-exchangeconditions. Trials preceding an exchangeperiod are shown on the abscissa, withproximity decreasing from left to right. Forall pigeons, both trial initiation (Figure 6) andchoice latencies (Figure 7) were longer in theearly than in the late trials preceding ex-change, and in many cases were a gradedfunction of trial position.

Choice patterns in delayed-exchange condi-tions were also analyzed in relation to trialposition and Equation 1. The closed circles inFigure 8 show the proportion of choices madefor the variable (MT) schedule as a function oftrial position preceding an exchange period.The leftmost point on the x-axis represents thefirst token earned (the point furthest from theupcoming exchange), and the rightmost pointthe trial immediately preceding exchange.The reference lines show the relative value ofthe MT (variable) alternative across trialpositions preceding an exchange period un-der the MT 1 s, 59 s (leftmost panels), MT 10 s,

50 s (center panels) and FT 30 s delayed-exchange conditions, computed with respectto exchange delays according to Equation 1(with D being the total average delay to theupcoming exchange period). To provide anexample of how relative value was computed,in the MT 1 s, 59 s condition the average delayto exchange from the start of a block of 11trials is 330 s. At this choice point, the values ofthe two alternatives are roughly similar. Theapproximate value of the adjusting delay is0.120 (computed with D 5 330) and the MTdelay 0.121 (computed with D 5 301 and D 5359, each occurring with p 5 0.5). Whenmaking a choice to earn the 11th token (thetoken directly preceding exchange), the valuesof the alternatives diverge, with the MT havinga value of 14.54 and the adjusting delay anapproximate value of 1.31. The relative valuewas then computed as the value of the MTalternative divided by the summed value ofboth alternatives. For the MT 1 s, 59 s and MT10 s, 50 s conditions, the relative value islargely equivalent until the trial most proximalto the upcoming exchange period, at whichpoint it shifts strongly in favor of the variable

Fig. 6. Mean latencies to respond on the center trial-initiation key as a function of trial position across successiveblocks of trials preceding an exchange period under delayed-exchange conditions. The leftmost point on the abscissaindicates the trial furthest from the upcoming exchange period. The dashed line in each plot indicates the comparablemean latencies under immediate-exchange conditions. Note the differently scaled y-axes.


(MT) schedule. This predicted upturn inpreference for the variable option in the trialpreceding exchange is consistent with thedata for 3 of the 4 subjects. With theexception of Pigeon 967 in the MT 1 s, 59 scondition, choice proportions were higheston the final trial preceding an exchangeperiod during the MT 1 s, 59 s and MT 10 s,50 s conditions, indicating strong preferencefor the variable schedule in close temporalproximity to an exchange. Choices during theFT 30 s schedule were largely undifferentiatedthroughout the 11 trials preceding exchange,which is consistent with the predictions ofEquation 1.

DISCUSSION

The current experiment shows that pigeonsare risk prone with respect to reinforcerdelays, preferring variable delays at values lessthan the mean reinforcer delay, a finding thatis consistent with previous nonhuman researchwith food reinforcers (e.g., Bateson & Kacel-nik, 1995, 1997; Cicerone, 1976; Herrnstein,1964; Killeen, 1968; Mazur, 1987; Rider, 1983).Additionally, risky-choice patterns were welldescribed by the hyperbolic decay model when

choices produced tokens immediately ex-changeable for food. According to Equation1, subjects should under immediate-exchangeconditions prefer the variable (MT) schedulebecause the 50% probability of receiving theshort reinforcement delay outweighs the valueof the adjusting reinforcer delay, even whenaccounting for the diminished value of therelatively longer MT delay (Mazur, 2004).More specifically, obtained indifference pointsshould be graded in relation to the short delayin the MT pair, such that lower indifferencepoints are expected when the short delays aresmallest. Our results confirm this: the pigeonsmore frequently chose the variable schedule,and obtained indifference points were orderedwith respect to the shortest delay in the MTschedule. Choice patterns were well describedby Equation 1 (holding K constant at 1) atboth the individual and group level. This, too,is consistent with prior research (Mazur, 1984,1986b, 1987). The fit was improved onlyslightly (95.2% vs. 95.0% of the varianceaccounted for) when K was allowed to varyacross subjects (accomplished by reducing thesum of the squared residuals using the Solverfunction in Microsoft ExcelH).

Fig. 7. Mean latencies to respond on the choice keys as a function of trial position across successive blocks of trialspreceding an exchange period under delayed-exchange conditions. The dashed line in each plot indicates thecomparable mean latencies under immediate-exchange conditions. Note the differently scaled y-axes.


Conditions were also designed to encompassfeatures common to most human risky choiceexperimentation. When subjects were re-quired to produce 11 tokens prior to exchangefor food, choice patterns differed markedlyfrom those seen under immediate-exchangeconditions. First, substantially higher indiffer-ence points were obtained, indicating some-times extreme risk aversion. Sensitivity to delaywas nevertheless demonstrated. First, indiffer-ence points were ordered with respect to theshortest delay in the MT distribution. Second,

there was marked preference for the variableoption on trials most proximal to exchangeperiods.

Preferences were systematically related toresponse latencies. In both immediate anddelayed exchange conditions, shorter latenciesto trial initiation and to choice were associatedwith risk-prone choice (lower indifferencepoints). Moreover, in delayed-exchange con-ditions, both trial initiation (Figure 6) andchoice latencies (Figure 7) were longer in theearly than in the late trials, a result that is

Fig. 8. Closed circles show the proportion of choices for the mixed-time (MT) delay as a function of trial positionacross a block of trials preceding an exchange period in the MT 1 s, 59 s condition (left panels), the MT 10 s, 50 scondition (center panels), and the FT 30 s condition (right panels). The reference lines show the relative value (V) of theMT alternative across trials, computed according to Equation 1 with K 5 1.


broadly consistent with the shift toward morerisk-prone choices in trials proximal to theupcoming exchange (Figure 8). Such tempo-ral regularities are consistent with thosereported under extended-chained schedulesof reinforcement (e.g., Bullock & Hackenberg,2006; Webbe & Malagodi, 1978), supporting aview of the delayed-exchange procedure as akind of token reinforcement schedule (seeHackenberg, 2009, for a review).

For none of the 4 subjects did the inclusionof an ITI have a discernible effect uponpreferences, in that indifference points stabi-lized close to those obtained in previousconditions using the same MT schedule. Thisfinding confirms Mazur’s (1988, 1989) dem-onstration that the ITI has only a minimaleffect upon choices compared to delaysbetween choice and reinforcement, and willnot be discussed further.

EXPERIMENT 2

In Experiment 1, delays to reinforcementwere discounted more sharply in the immedi-ate-exchange than in the delayed-exchangeconditions. The reduced delay discountingwhen reinforcement is provided after a fixedperiod may at least partially account for themixed results of previous human choiceresearch using points later exchangeable formoney at the end of the experimental session.This is the subject of Experiment 2.

Experiment 2 was designed to mimic criticalfeatures of Experiment 1 but with humansinstead of pigeons and 30-s video clips insteadof food. Participants chose between variableand adjusting delays to reinforcement underimmediate-exchange and delayed-exchangeconditions. Like Experiment 1, the currentexperiment included forced-choice trials andsteady-state conditions that remained in effectuntil a predetermined stability criterion wasmet. If procedural variables are indeed thereason for the behavioral disparities acrossspecies, then humans’ choices—like theirpigeon counterparts in Experiment 1—shouldfavor the variable schedule more strongly inimmediate-exchange conditions than in de-layed-exchange conditions. Moreover, if thevideo clips function as effective reinforcers,one might further expect Equation 1 todescribe the choice patterns.

METHOD

Subjects

One male and 4 female undergraduatestudents between the ages of 18 and 22participated. They were recruited via adver-tisements in a local university newspaper andfliers posted around the campus, and wereselected for participation based on their dailyavailability, inexperience with behavior analy-sis, and high interest in watching televisionshows. Subjects were paid a flat rate of $3.00per day, with an additional $3.50 per hourpaid contingent upon completion of the study.The average overall earnings were $5.99 perhour (range, $5.90 to $6.08). Subjects were notpaid until participation was completed. Priorto the beginning of the study, subjects signedan informed consent form and were told thatthey would be expected to participate 5 daysper week for approximately 2 months.

Apparatus

Sessions were conducted in a closed exper-imental room measuring 2.7 m wide and 1.8 mlong. Subjects were seated in front of aGateway PC computer on which the experi-mental contingencies were arranged and datarecorded via Microsoft Visual BasicH software.The computer was programmed to displaythree square response buttons (each 5 cm by 5cm) that could be operated by positioning thecursor over one and clicking it with thecomputer mouse. Twelve circles located 3 cmabove the response buttons served as tokensand were illuminated red when earned. Tokenillumination and darkening was accompaniedby a 0.1-s beep. Reinforcement consisted ofvideo clips from preferred television programsoperated by Windows Media PlayerH and weredisplayed on the full screen of the computerduring reinforcement periods.

Procedure

Subjects were exposed to two sessions perday, 5 days per week. Upon arrival, eachsubject was escorted to the experimentalroom, leaving behind possessions (e.g., watch-es, cell phones, book bags) in a differentroom. At the start of each session a subject wasasked to select a television show to watch froma computer-displayed list of the following eightprograms: Friends, Will and Grace, Family Guy,Wallace and Gromit, Sports Bloopers, Looney Tunes,


The Simpsons, and Seinfeld. Many of the pro-gram choices were series with multiple epi-sodes that were cycled through in sequence.Therefore, if a subject so chose, episodes of asingle program could be viewed in series acrosssuccessive sessions. A session consisted of asingle full-length episode of a program (ap-proximately 22 min, with commercials omit-ted) divided into 44 equivalent segmentsapproximately 30 s in duration.

Once the program had been selected, thefollowing instructions were displayed on thescreen: ‘‘Use the mouse to earn access tovideos. You will need to use only the mouse forthis part of the experiment. When you areready to begin, click the ‘begin’ buttonbelow.’’ At the end of each two-session visit,subjects rated the two recently viewed videoson a 1-5 point scale ranging from ‘‘VeryGood’’ to ‘‘Very Bad’’. Subjects also wereasked to generally evaluate the sessions bytyping into the computer an open-endedverbal response.

The start of each trial was marked by thesimultaneous presentation of three squareresponse buttons, centered vertically andhorizontally on the screen. Only the centerbutton was activated and illuminated; onemouse click on this trial-initiation buttonwould then deactivate this button and allowaccess to the side choice keys. A single click oneither key would initiate the scheduled delayand deactivate the nonchosen button. Thechosen button remained illuminated through-out the delays. When the delay had elapsed,one token light was illuminated simultaneous-ly with a 1-s beep. In immediate-exchangeconditions, the exchange period (signaled by ared center button) occurred after a singletoken presentation. A single click on theexchange button presented full-screen accessto the 30-s video clip. In delayed-exchangeconditions, the exchange period was present-ed when 11 tokens had been earned. Each ofthe 11 tokens could be exchanged in immedi-ate succession for eleven 30-s video clips bymaking one exchange response per video clip.The choice screen was presented again aftereach reinforcer, permitting additional ex-change (if tokens remained) or the start of anew trial (if no tokens remained).

Prior to the start of the experiment, a delaysensitivity probe was implemented to assesscontrol by delay to video reinforcement. In

these sessions, subjects were given repeatedchoices between a 1-s and a 15-s delay to theonset of a 30-s video clip for at least twosessions of 44 trials before moving to theexperiment proper. Of 6 participants tested,one did not consistently prefer the shorterdelay to reinforcement and was terminatedfrom participation in the study.

As in Experiment 1, MT distribution andexchange delay were manipulated systemati-cally across conditions. The average IRI washeld constant at 15 s, with standard MT delaysof (a) 1 s, 29 s; (b) 10 s, 20 s; and (c) 15 s, 15 s(FT 15 s). At the beginning of each condition,the adjusting delay started at the average IRI of15 s. Each selection of the MT alternativedecreased the adjusting delay on the followingtrial by 10%, while each selection of theadjusting delay increased that delay by 10%on the subsequent trial. The adjusting delayhad a lower limit of 1 s and an upper limit of60 s. Conditions lasted a minimum of foursessions and were deemed stable via visualinspection of trend and bounce. Table 2 showsthe sequence and number of conditions perparticipant. Two participants (247 and 250)voluntarily left the study early and did notcomplete all of the planned experimentalconditions.

RESULTS

Indifference points were determined byaveraging the obtained adjusting delays overthe last three stable data points per condition.Figure 9 displays these indifference points forall participants, with the resulting pooled datafitted to Equation 1 with parameter K fixed at1 (the same value used in Experiment 1). Datafrom initial and replication conditions wereaveraged, providing one indifference pointper participant per condition. Equation 1provided a good fit to the pooled data, with83.5% of the variance accounted for.

Because most of the participants completedconditions at only two different MT delays,individual curves were not plotted. Perfor-mances were generally similar across individu-als, however. Indifference points were orderedwith respect to relative reinforcer immediacy,and there was little overlap in the distributionof points comprising the means. Indifferencepoints ranged between 1.1 s and 2.6 s (M 51.94 s) under MT 1 s, 29 s, from 9.2 s to 13.5 s(M 5 11.0 s) under MT 10 s, 20 s, and from


12.9 to 17.2 (M 5 14.4 s) under the FT 15-scontrol condition. The latter result indicates ageneral lack of bias for either alternative.

Mean indifference points obtained duringthe immediate and delayed exchange condi-tions are compared in Figure 10, with errorbars displayed for conditions that were repli-cated (note the higher y-axis scale for Partic-ipant 245). As in Experiment 1, the delayed-exchange conditions resulted in less discount-ing of reinforcer delay, as evidenced by thehigher indifference points relative to thoseobtained during immediate-exchange condi-tions. This was particularly evident when theMT values contained relatively long delays toreinforcement (i.e., the MT 10 s, 20 scondition), with mean indifference points of28.2 s under delayed exchange and 11 s underimmediate exchange. There was also evidenceof delay sensitivity under delayed exchange. Inall 3 participants who completed more thanone delayed exchange condition, indifferencepoints were higher under conditions withrelatively longer short delays (e.g., MT 10 s,20 s versus MT 1 s, 29 s.)

As in Experiment 1, latencies to trialinitiation, choice, and token exchange re-sponses were also examined as a function ofthe different MT delay conditions. Figures 11and 12 display this result during the immedi-ate and delayed exchange conditions, respec-tively, averaged across subjects. In immediate-exchange conditions (Figure 11), latencies totrial initiation and exchange both increased asa function of the short delay in the MT pair,but choice latencies were largely unaffected byMT schedule. As in Experiment 1, whenlatencies changed across conditions they werepositively related to obtained indifferencepoints, with shorter latencies occurring inconditions in which shorter adjusting delayswere obtained. Response latencies in the

delayed exchange conditions were largelyundifferentiated across conditions, but weresomewhat longer for exchange responses.

DISCUSSION

The current experiment brought into closeralignment the choice procedures used withhuman and nonhuman subjects. First, partic-ipants were exposed to experimental contin-gencies repeatedly over trials and acrosssessions, rather than the more typical practiceof ‘‘one-shot’’ choices. Also, unlike priorresearch with hypothetical outcomes, partici-pants earned access to actual consequences:video clips of favorite TV programs. Byproviding immediately consumable reinforce-ment in a repeated-trials format, it washypothesized that participants would maketheir choices based on the programmed delays

Table 2

Sequence and number of sessions per condition (in parentheses) for each participant inExperiment 2.

Condition 245 247 248 249 250

FT 15 s Immediate 1(4) 1(4) 1(4) 1(4) 1(18)7(12) 6(12)

MT 1 s, 29 s Immediate 2(10) 2(4) 2(10) 2(4) 2(10)6(10) 6(16) 7(12)

MT 10 s, 20 s Immediate 4(8) 4(10) 4(6)MT 1 s, 29 s Delayed 3(19) 3(4) 3(6) 3(6)MT 10 s, 20 s Delayed 5(10) 5(14) 5(6)

Fig. 9. Mean indifference points as a function of themixed-time (MT) or fixed-time (FT) delay condition for allsubjects in Experiment 2. The curve displays predictedindifference points from Equation 1 with K 5 1.


to reinforcement rather than the overall-session rate of reinforcement. Our resultsconfirm this: During immediate exchangeconditions, choices were sensitive to delay,resembling that of nonhuman subjects (in-cluding those in Experiment 1). Indifferencepoints conformed well to the predictions ofEquation 1, even when employing the samediscounting parameter value as used with food-deprived pigeon subjects. The delayed-ex-change conditions produced elevated indiffer-ence points, indicating greater risk aversion,more like that often seen in laboratory risky-choice procedures with humans.

As in Experiment 1, latencies to trialinitiation, choice, and token exchange re-sponses were examined as a function of thedifferent MT delay conditions. A similarpattern was evident under immediate-ex-change conditions for the humans in thisexperiment as for the pigeons in Experiment1: latencies to trial initiation and exchange

Fig. 10. Mean indifference points for individual subjects from immediate exchange (black bars) and delayedexchange (grey bars) conditions as a function of the mixed-time (MT) or fixed-time (FT) delay, including standarddeviations. Note the extended y-axis for Subject 245.

Fig. 11. Mean latencies to respond on choice, ex-change, and trial initiation keys as a function of the mixed-time (MT) or fixed-time (FT) delay, averaged acrosssubjects, in immediate exchange conditions. Error barsindicate across-subject standard deviations.


increased as obtained adjusting delays for thatMT schedule increased.

The use of video reinforcement also extendsprior research with humans (e.g., Navarick,1996; 1998; Hackenberg & Pietras, 2000; Loceyet al., 2009) and demonstrates the utility ofthis reinforcer type for use in laboratoryresearch. Across the range of conditionsstudied here, video reinforcers shared impor-tant functional characteristics with more com-mon consumable reinforcers (e.g., food).Although many questions remain concerningits relative reinforcer efficacy, the presentresults suggest that video reinforcers canfunction as an effective tool in laboratoryresearch with humans.

GENERAL DISCUSSION

The present study compared the choicepatterns of pigeons and humans in a risky-choice context. A major objective was to bringhuman and pigeon procedures into closeralignment, facilitating species comparisons. Tothat end, pigeons and humans were exposedto analogous procedures that included forced-choice trials, steady-state analyses, and con-sumable-type reinforcement across two differ-ent economic contexts: immediate-exchange,most analogous to prior research with nonhu-mans, and delayed-exchange, most analogousto prior research with humans. Thus, ifprocedural differences are responsible forprior species differences, then choice patterns

of both species should be more risk-proneunder immediate-exchange than under de-layed-exchange conditions. Moreover, if choic-es are sensitive to delay, then indifferencepoints should be ordered with respect toreinforcer immediacy. Both of these resultswere confirmed.

Discounting by reinforcer delay was seenboth with pigeons (Experiment 1) and hu-mans (Experiment 2). This was reflected inboth (a) the greater number of choices madefor the variable (MT) schedule under theimmediate-exchange conditions (as indexedby lower indifference points); and (b) theordering of indifference points with respect torelative reinforcer immediacy. Both of theseresults are in accord with previous research(e.g., Bateson & Kacelnik, 1995; Gibbon, et al.,1988; Mazur, 1987; 2006). The results are alsobroadly consistent with Equation 1, a model ofhyperbolic reinforcer discounting. The modelaccounted for 95% and 83.5% of the overallvariance in the pigeon and human data sets,respectively. For individual pigeon subjects,using K values of 1.0 provided reasonable fitsto Equation 1, with the percentages of varianceaccounted for ranging from 91.2% to 97.3%across subjects. There is ample precedent forallowing K to vary (e.g., Mazur, 1986b, 1987;Myerson & Green, 1995); however, doing soproduced only moderately higher percentagesranging from 91.6% to 97.3% for pigeons and83.5% to 84.5% for humans.

Although we did not have enough datapoints to provide informative best-fit K valuesfor individual human data, the pooled K valueswere higher than those frequently reported forhuman choices with hypothetical outcomes(Lagorio & Madden, 2005; Myerson & Green,1995; Petry & Casarella, 1999). Attaininghigher K values is likely due to the consum-able-type reinforcers used in the current study.This type of reinforcer allowed for theassessment of preferences within a time framemore comparable to that consistently usedwith nonhumans: actual reinforcers delayed byseconds or minutes rather than reinforcersthat are either hypothetical or delayed by daysor years.

More direct comparisons across species areprovided in Figure 13, which shows obtainedindifference points for pigeons and humans inthe present study. To facilitate comparisons,indifference points are plotted as a function of

Fig. 12. Mean latencies to respond on choice, ex-change, and trial initiation keys as a function of the mixed-time (MT) or fixed-time (FT) delay, averaged acrosssubjects, in delayed exchange conditions. Error barsindicate across-subject standard deviations.


the predicted equivalent adjusting delaysprovided by Equation 1. The data for bothspecies are ordered according to the predic-tions of the hyperbolic-decay model. On thewhole, Equation 1 provided a good descriptionof the data for both pigeons and humans,suggesting similarities in the way delayedreinforcers are discounted for different spe-cies, whether those reinforcers are food forpigeons or videos for humans.

Although broadly consistent with Equation1, the results are also consistent with otherdiscounting models. Both exponential andlinear models provided generally comparablefits. Given the generally good fits overall, moreprecise model testing must await additionalresearch. The present analysis suggests, how-ever, that the methods employed here are wellsuited to quantitative cross-species compari-sons in delay discounting.

In delayed-exchange conditions, both pi-geons and humans favored the adjustingalternative (producing higher indifferencepoints) to a far greater degree than whentokens were exchanged immediately. As in theimmediate-exchange conditions, indifferencepoints increased with relative reinforcer delay,but the indifference points were uniformlyhigher, indicating shallower discounting.

These differences between immediate anddelayed exchange show that risky choicedepends in part on the economic context.Reinforcers were discounted more sharply in

contexts permitting immediate consumptionthan in contexts requiring accumulation ofreinforcers for later exchange. This basicresult confirms previous research (e.g., Hyten,Madden, & Field, 1994; Jackson & Hacken-berg, 1996), but to date, the relevant data arefrom comparisons across studies. The presentstudy is noteworthy in that it provides directcomparisons of pigeons and humans in bothtypes of economic context. That the economiccontext produced generally similar results forpigeons and humans attests to the majorinfluence of this variable.

The latency measures revealed some simi-larities across species, but differed in quanti-tative detail. For pigeons, trial-initiation andchoice latencies varied with relative reinforcerdelay under both immediate and delayedexchange, with shorter overall latencies occur-ring in conditions with lower indifferencepoints. For humans, trial-initiation latencieswere sensitive to reinforcer delay underimmediate but not delayed exchange, withchoice latencies relatively undifferentiatedacross conditions. In many cases, the latencymeasures were inversely related to preference,as might be expected from a unitary concep-tion of reinforcer value. The overall pattern ofresults shows how local measures of behaviorcan be usefully brought to bear on concep-tions of reinforcer value.

In sum, the present results suggest thatpigeons’ and humans’ choices are influencedin similar ways by reinforcer delay. The generalcomparability of results seen with pigeons andhumans may be due to the use of ‘‘consumable-type’’ reinforcement with humans. Other stud-ies employing such consumable-type reinforc-ers with humans have also reported preferencesmore in line with the delay sensitivity seen withnonhumans (e.g., Forzano & Logue, 1994;Hackenberg & Pietras, 2000; Locey et al.,2009; Millar & Navarick, 1984). Along withexposure to forced-choice trials and repeatedexposure to the contingencies across multipledaily sessions, the use of consumable-typereinforcers brought the procedures into great-er alignment with those typically used withother animals. The present results join with agrowing body of findings showing that previ-ously reported differences between humansand other animals may be at least partlyattributed to procedural features. Only withprocedures matched on important functional

Fig. 13. Individual-subject indifference points for pi-geon (closed circle) and human (open square) subjectsunder immediate-exchange conditions as a function of thepredicted delay (D) from Equation 1.


characteristics can genuine species differencesbe separated from procedural differences.

REFERENCES

Bateson, M., & Kacelnik, A. (1995). Preferences for fixedand variable food sources: Variability in amount anddelay. Journal of the Experimental Analysis of Behavior, 63,313–329.

Bateson, M., & Kacelnik, A. (1997). Starlings’ preferencesfor predictable and unpredictable delays to food.Animal Behavior, 53, 1129–1142.

Bullock, C. E., & Hackenberg, T. D. (2006). Second-orderschedules of token reinforcement with pigeons:Implications for unit price. Journal of the ExperimentalAnalysis of Behavior, 85, 95–106.

Case, D. A., Nichols, P., & Fantino, E. (1995). Pigeons’preference for variable-interval water reinforcementunder widely varied water budgets. Journal of theExperimental Analysis of Behavior, 64, 299–311.

Cicerone, R. A. (1976). Preference for mixed versusconstant delay of reinforcement. Journal of the Exper-imental Analysis of Behavior, 25, 257–261.

Davison, M. C. (1969). Preference for mixed-intervalversus fixed-interval schedules. Journal of the Experi-mental Analysis of Behavior, 12, 247–252.

Davison, M. C. (1972). Preference for mixed-intervalversus fixed-interval schedules: Number of compo-nent intervals. Journal of the Experimental Analysis ofBehavior, 17, 169–176.

Fantino, E. (1967). Preference for mixed- versus fixed-ratioschedules. Journal of the Experimental Analysis ofBehavior, 10, 35–43.

Forzano, L. B., & Logue, A. W. (1994). Self-control in adulthumans: Comparison of qualitatively different rein-forcers. Learning and Motivation, 25, 65–82.

Frankel, P. W., & vom Saal, W. (1976). Preference betweenfixed-interval and variable-interval schedules of rein-forcement: Separate roles of temporal scaling andpredictability. Animal Learning and Behavior, 4, 71–76.

Gibbon, J., Church, R. M., Fairhurst, S., & Kacelnik, A.(1988). Scalar expectancy theory and choice betweendelayed rewards. Psychological Review, 95, 102–114.

Hackenberg, T. D. (2009). Token reinforcement: A reviewand analysis. Journal of the Experimental Analysis ofBehavior, 91, 257–286.

Hackenberg, T. D., & Pietras, C. J. (2000). Video access asa reinforcer in a self-control paradigm: A method andsome data. Experimental Analysis of Human BehaviorBulletin, 18, 1–5.

Herrnstein, R. J. (1964). Aperiodicity as a factor in choice.Journal of the Experimental Analysis of Behavior, 7,179–182.

Hursh, S. R., & Fantino, E. (1973). Relative delay ofreinforcement and choice. Journal of the ExperimentalAnalysis of Behavior, 19, 437–450.

Hyten, C., Madden, G. J., & Field, D. P. (1994). Exchangedelays and impulsive choice in adult humans. Journalof the Experimental Analysis of Behavior, 62, 225–233.

Jackson, K., & Hackenberg, T. D. (1996). Token reinforce-ment, choice, and self-control in pigeons. Journal of theExperimental Analysis of Behavior, 66, 29–49.

Kacelnik, A., & Bateson, M. (1996). Risk theories—theeffects of variance on foraging decisions. AmericanZoologist, 36, 402–434.

Kendall, S. B. (1989). Risk-taking behavior of pigeons in aclosed economy. Psychological Record, 39, 211–220.

Killeen, P. (1968). On the measurement of reinforcementfrequency in the study of preference. Journal of theExperimental Analysis of Behavior, 11, 263–269.

Kohn, A., Kohn, W. K., & Staddon, J. E. R. (1992).Preferences for constant duration delays and constantsized rewards in human subjects. Behavioural Processes,26, 125–142.

Lagorio, C. H., & Madden, G. J. (2005). Delay discountingof real and hypothetical rewards III: Steady-stateassessments, forced-choice trials, and all real rewards.Behavioural Processes, 69, 173–187.

Locey, M. L., Pietras, C. J., & Hackenberg, T. D. (2009).Human risky choice: Delay sensitivity depends onreinforcer type. Journal of Experimental Psychology:Animal Behavior Processes, 35, 15–22.

Logan, F. A. (1965). Decision making by rats: Uncertainoutcome choices. Journal of Comparative and Physiolog-ical Psychology, 59, 246–251.

Mazur, J. E. (1984). Tests of an equivalence rule for fixedand variable reinforcer delays. Journal of ExperimentalPsychology: Animal Behavior Processes, 10, 426–436.

Mazur, J. E. (1986a). Choice between single and multipledelayed reinforcers. Journal of the Experimental Analysisof Behavior, 46, 67–77.

Mazur, J. E. (1986b). Fixed and variable ratios and delays:Further tests of an equivalence rule. Journal ofExperimental Psychology: Animal Behavior Processes, 12,116–124.

Mazur, J. E. (1987). An adjusting procedure for studyingdelayed reinforcement. In M. L. Commons, J. E.Mazur, J. A. Nevin, & H. Rachlin (Eds.), Quantitativeanalyses of behavior, Vol. 5. Effects of delay and interveningevents on reinforcement value (pp. 55–73). Hillsdale, NJ:Erlbaum.

Mazur, J. E. (1988). Estimation of indifference points withan adjusting delay procedure. Journal of the Experimen-tal Analysis of Behavior, 49, 37–47.

Mazur, J. E. (1989). Theories of probabilistic reinforce-ment. Journal of the Experimental Analysis of Behavior, 51,87–99.

Mazur, J. E. (2004). Risky choice: Selecting betweencertain and uncertain outcomes. Behavior AnalystToday, 5, 190–203.

Mazur, J. E. (2006). Mathematical models and theexperimental analysis of behavior. Journal of theExperimental Analysis of Behavior, 85, 275–291.

Mazur, J. E. (2007). Rats’ choices between one and twodelayed reinforcers. Learning and Behavior, 35,169–176.

Millar, A., & Navarick, D. J. (1984). Self-control and choicein humans: effects of video game playing as a positivereinforcer. Learning and Motivation, 15, 203–218.

Myerson, J., & Green, L. (1995). Discounting of delayedrewards: Models of individual choice. Journal of theExperimental Analysis of Behavior, 64, 263–276.

Navarick, D. J. (1996). Choice in humans: Techniques forenhancing sensitivity to reinforcement immediacy.The Psychological Record, 46, 539–554.

Navarick, D. J. (1998). Impulsive choice in adults: Howconsistent are individual differences? The PsychologicalRecord, 48, 665–674.

Navarick, D. J., & Fantino, E. (1975). Stochastic transitivityand the unidemensional control of choice. Learningand Motivation, 6, 179–201.


O’Daly, M., Case, D. A., & Fantino, E. (2006). Influence ofbudget and reinforcement location on risk-sensitivepreference. Behavioural Processes, 73, 125–135.

Orduna, V., & Bouzas, A. (2004). Energy budget versustemporal discounting as determinants of preferencein risky choice. Behavioural Processes, 67, 147–156.

Petry, N. M., & Casarella, T. (1999). Excessive dis-counting of delayed rewards in substance abuserswith gambling problems. Drug and Alcohol Dependence,56, 25–32.

Rider, D. (1983). Preference for mixed versus constantdelays of reinforcement: Effect of probability of theshort, mixed delay. Journal of the Experimental Analysis ofBehavior, 39, 257–266.

Rodriguez, M. L., & Logue, A. W. (1988). Adjusting delayto reinforcement: Comparing choice in pigeons andhumans. Journal of Experimental Psychology: AnimalBehavior Processes, 14, 105–117.

Sherman, J. A., & Thomas, J. R. (1968). Some factorscontrolling preference between fixed-ratio and vari-able-ratio schedules of reinforcement. Journal of theExperimental Analysis of Behavior, 11, 689–702.

Webbe, F. M., & Malagodi, E. F. (1978). Second-orderschedules of token reinforcement: Comparisons ofperformance under fixed-ratio and variable-ratioexchange schedules. Journal of the Experimental Analysisof Behavior, 30, 219–224.

Weiner, H. (1966). Preference and switching under ratiocontingencies with humans. Psychological Reports, 18,239–246.

Zabludoff, S. D., Wecker, J., & Caraco, T. (1988). Foragingchoice in laboratory rats: Constant vs. variable delay.Behavioural Processes, 16, 95–110.

Received: June 10, 2009Final Acceptance: September 2, 2009


Documents

· reinforcer (i.e., self-control), a result typically observed with humans. In other words, when the procedures were more typical of nonhu-man self-control procedures—differential