Listed below are the SQAB abstracts for Presentations, Poster Presentations, and Preeminent Tutorials.

Friday, May 23

William M. Baum
University of California-Davis

Molar and Molecular Views of Choice

Choice is ubiquitous. To study behavior is to study choice, because every situation entails more than one behavioral alternative. The molecular and molar views of behavior deal with this fundamental principle differently. In the molecular view, choices consisting of discrete responses occur at moments in time. Such responses are considered concrete, and extended patterns, such as a response rate or relative response rate, are considered to be abstractions created out of the concrete discrete responses. In the molar view, extended patterns of activity are considered concrete and discrete responses are considered to be abstractions. These differences lead the two views to deal differently with another fundamental principle, that present behavior always occurs in the context of past events. Because the molecular view seeks to explain momentary choices, it requires immediate causes, which represent the past in the present. The molecular view thus forces explanations to rely on hypothetical constructs, such as response strength and expectancy. The molar view incorporates the past as inhering in the extendedness of activity patterns. Every activity is composed of less extended activities and constitutes part of a more extended activity. Molar explanations require no hypothetical constructs because extended activity patterns are always directly measurable.

Michael Davison
University of Auckland

Interresponse Times and the Structure of Choice

Time allocation has been a widely-used measure of choice, and a number of researchers have argued that time allocation is the fundamental process in choice. I shall present a series of new analyses of recently published research that casts doubt on the notion that concurrent- schedule choice arises from discriminable bouts of time spent emitting defined operants, and that this process leads to the close correspondence between response and time measures of preference. I shall show that most time allocation arises not from emitting the defined response, but from emitting and allocating time to other, undefined, responses. I shall show how consistent distributions of time spent emitting undefined responses may be reconstructed using log-normal analyses. Understanding choice may require understanding the transitions between the large number of defined and undefined behaviors available even in simply choice situations.

James E. Mazur
Southern Connecticut State University

Concurrent-chains schedules as tools to test theories of choice

Several different models of concurrent-chains performance have been developed, including delay- reduction theory, the contextual-choice model, and the hyperbolic value-added model. For many typical concurrent-chains schedules, these models make similar predictions, and it is therefore difficult to distinguish between them. However, for some specific conditions the models make distinctly different predictions, and these situations can be used to provide critical tests of the models. Data from experiments where initial-link durations were varied, where terminal-link durations were varied, and where different delay-amount combinations were used will be examined to compare the predictions of the three models. Such comparisons can provide useful information about the processes that govern choice behavior.

Barbara Mellers
University of California-Berkeley

Predicting Emotional Reactions to the Outcomes of Choice

After a decision, we face the consequences. This talk examines the extent to which our affective reactions of pleasure and pain associated with a given outcome depend on salient reference points and beliefs about possible outcomes. We present a theory of choice and a related theory of the pleasure that of outcomes that follow from choice. These theories imply that certain systematic biases in beliefs (overconfidence and hindsight) can decrease the pleasure that follows from choice. Experimental results confirm those predictions.

Gordon R. Foxall
Cardiff University

The behavioral economics of consumer brand choice

Matching theory predicts choices on concurrent variable ratio schedules will show maximization via exclusive choice of the richest schedule. In earlier papers we have reported patterns of brand selection data from individual consumers which show that consumer behavior exhibits both matching and maximization. In this paper we report the results of a study of 80 consumers' brand selections for 8 product categories which indicate that the patterns identified in our initial research can be generalized. We discuss the implications of our findings for research in the behavioral economics of consumption in marketing-oriented economies and for the issue of what and how consumers maximize.

Howard Rachlin
State University of New York at Stony Brook

Cooperation by Pigeons Versus Tit-For-Tat In The Prisoner's Dilemma

The prisoner's dilemma becomes a self-control problem when playing against a tit-for-tat strategy. But the kind of self-control required is not just a simple choice between a larger- later and smaller-sooner reward (simple ambivalence). In the typical prisoner's dilemma versus tit-for-tat, a larger overall average reward can be obtained only by repeatedly choosing and consuming the smaller of two immediately available rewards. The conflict in the prisoner's dilemma is better described as an opposition between global and local maximizing (complex ambivalence) than as a simple choice among discounted rewards. This complexity has made it difficult to train nonhumans to exhibit self-control in prisoner's dilemma situations against tit-for-tat. These experiments show, however, that when differential rewards are clearly signaled and when response feedback is provided, pigeons can learn to repeatedly choose the smaller of two immediate rewards in order to maximize overall reward. In other words, pigeons can learn to "cooperate" in the prisoner's dilemma against tit-for-tat. Pigeons also learn to cooperate against tit-for-tat when the normal prisoner's dilemma requirement of reward consumption after each choice is removed by reinforcing only a small fraction of choices (in an operant choice procedure).

Ron Weisman
Queen's University

Contributions of Behavior Analysis to Understanding the Evolution of Pitch Perception

Absolute pitch perception (AP) requires the sorting of pitches into classes without an external frequency referent. Highly accurate AP is rare in humans, much-prized by musicians, but of no demonstrable value to musicians or other humans. Field experiments that play altered song to territorial male songbirds of several species have repeatedly shown that shifting the frequency (and hence the perceived pitch) of a conspecific playback song out of the frequency range of conspecific songs reduced males' response to the level observed with heterospecific song. Hence, AP is demonstrably important to songbirds (who must recognize the songs of conspecifics to defend territories and reproduce). My colleagues and I developed operant discrimination procedures for studying AP in the laboratory. When zebra finches (a songbird species) and humans were trained to sort contiguous pitches into 3 or 8 ranges, based on correlations between tones in each range and reinforcement and nonreinforcement in go/no-go discriminations, (a) songbirds discriminated 3 ranges to a much higher standard than humans, and (b) songbirds discriminated 8 ranges with precision whereas humans acquired only a crude discrimination of the lowest and highest of 8 ranges. It is possible that the mediocre AP of most humans (9,999 out of 10,000) has a uniquely human origin in the acquisition of language or music (as many psychologists suppose). Alternatively, human inferiority and songbird superiority at AP may reflect more general evolutionary trends among mammals and birds, respectively. We have used 3- and 8-range discriminations to test another mammalian species (rats), another songbird species (white-throated sparrows) and nonsongbird avian species (budgerigars). Rats do no better than humans in these quantitative pitch range tests of AP, whereas other songbirds and nonsongbirds sort pitches into ranges with about same high accuracy as zebra finches.

Julian C. Leslie
University of Ulster

Drug effects on extinction of operant behavior in mice: Implications for theoretical accounts of extinction

Using a procedure previously shown to be sensitive to chlordiazepoxide (CDP) in rats, C57/Black 6 male mice were trained in daily sessions to lever press for food on a discrete- trial FR schedule. In the first series of experiments, once FR behavior had stabilized drugs were administered prior to the last 2 training days and prior to all subsequent extinction sessions. As had been previously found in rats, CDP (15 mg/kg) tended to facilitate the extinction process. However, this effect only occurred after several days of drug administration and thus is not attributable to possible sedative effects. Two novel compounds, developed by Merck Neuroscience with subtype- selective benzodiazepine partial agonist actions at the GABA receptor complex, had a similar pattern of effects to those of CDP: Cpd A had a dose-related effect of selectively facilitating extinction in similar fashion to CDP, and Cpd B had an effect that was greater than that of CDP. In subsequent experiments, we further established that CDP has this effect even if only administered late in a series of extinction sessions, that it also enhances subsequent reacquisition of food-reinforced behavior in mice, and that this specific effect of CDP can also be seen when the rate of extinction is varied by changing the number of trials per extinction session or by changing the fixed ratio value. Various molecular analyses of the extinction process and the effects of CDP upon the process will be presented. These findings will be related to descriptive and theoretical accounts of operant extinction.

Raymond C. Pitts & Stacy Febbo
University of North Carolina at Wilmington

Quantitative Analyses of Stimulant Effects on "Self-Control" Choices: Implications for Elucidating Behavioral Mechanisms of Drug Action

The purpose of the present research was to use quantitative analyses to identify "behavioral mechanisms" involved in the effects of stimulant drugs on choice in a "self-control" procedure. Identification of behavioral mechanisms of drug action is complicated by the fact that behavior nearly always is multiply determined. For example, patterns of behavior maintained under self-control procedures involve at least the combined effects of reinforcement amount and reinforcement delay. Thus, it is difficult to determine whether a given drug effect under these conditions reflects a change in control by reinforcement delay, reinforcement amount, or both. The current approach involved applying a logarithmic form of the generalized the matching law; drug-induced changes in derived parameters for sensitivity to reinforcement delay and sensitivity to reinforcement amount were used to infer behavioral mechanisms of drug action. Pigeons were exposed to a procedure in which the delay to a larger reinforcer was manipulated within experimental sessions. A concurrent- chains procedure was employed. The initial link consisted of concurrent random-interval (RI) schedules. In the terminal links, one alternative provided 1-s access to food (the smaller reinforcer) and the other alternative provided 4-s access to food (the larger reinforcer). The signaled delay to the smaller reinforcer always was 2-s, whereas the signaled delay to the larger reinforcer was increased from 2-s to 40-s within each session, across 10-min blocks. That is, for the first 10-min of each session, the delay to the larger reinforcer was 2-s and for the next 10-min, it was 10-s, and so on. Under this procedure, relatively reliable delay-discount functions were obtained within experimental sessions for all birds. Establishing consistent within-session discount functions allowed us to examine effects of several doses of the stimulant drugs methamphetamine and methylphenidate on the entire function. In general, intermediate doses of both methamphetamine and methylphenidate increased preference for the larger more delayed reinforcer. Quantitative analyses indicated that, in many cases, both of these drugs decreased sensitivity to reinforcement delay. In a few instances, concomitant decreases in sensitivity to reinforcement amount also occurred. Although other interpretations certainly are possible, these results suggest that a reduced sensitivity to reinforcement delay, and perhaps a reduced sensitivity to reinforcement amount, may be important behavioral mechanisms of the effects of stimulants on self- control choices.

Leonard Green & Joel Myerson
Washington University

Discounting of Delayed and Probabilistic Rewards: Processes and Traits

Previous research has shown that the same form of hyperbola-like function provides good fits to individual discounting data obtained with both delayed and probabilistic rewards. Nevertheless, certain variables (e.g., amount of reward) differentially affect temporal and probability discounting. To provide further tests of hypotheses that temporal and probability discounting reflect the same basic mechanism or that individual differences in the two types of discounting reflect the same basic trait, we obtained data from two large samples (Ns > 100). Consistent with previous findings, a hyperbola- like discounting function provided good fits to individual temporal and probability discounting data, but amount of reward had opposite effects on the degree of discounting delayed and probabilistic rewards. Moreover, the nonlinear scaling parameter of the hyperbola-like function was larger for larger probabilistic rewards, but did not vary with the amount of delayed reward. Taken together, these findings suggest that despite the similar form of the temporal and probability discounting functions, separate processes may underlie the discounting of delayed and probabilistic rewards. Finally, weak to moderate positive correlations were observed between the discounting of delayed and probabilistic rewards. This finding is inconsistent with the notion of an "impulsiveness" trait that links an inability to delay gratification with a tendency to gamble and take risks.

K. Geoffrey White & Ee Lin Ong
University of Otago

Amount-dependent temporal discounting?

A popular account of preference for small immediate reinforcers over large delayed reinforcers is that reinforcer value is increasingly discounted as the delay is lengthened. Further increasing the delay to both reinforcers results in a reversal of preference. This result suggests that temporal discounting functions are hyperbolic and cannot be exponential. Discounting functions can be exponential, however, if the rate of discounting is slower for larger amounts, as indicated by studies with humans using titration procedures. Grace (1999) found no evidence for amount- dependent temporal discounting in pigeons. Using a modified version of his concurrent chains procedure, we found that choice was more sensitive to ratios of delays when the reinforcers were both large, compared to when they were both small. That is, the concurrent discrimination between delays was greater for larger reinforcers. This result is inconsistent with a slower rate of discounting for larger reinforcers. We report further studies that explore the inconsistent results from concurrent chains procedures in pigeons versus titration procedures in humans. An extension of the delay- reduction hypothesis, in which terminal-link waiting time is amplified by reinforcer amount, predicts our results.

Saturday Morning, May 24

Charles Shimp
University of Utah

Categorizing spatial and temporal configurations

The development of research on categorizing stimuli in non human animals parallels, with a lag of several decades, the development of Wittgenstein's views on the nature of knowledge. In particular, both began with a focus on logical relations among a small number of features or dimensions: Categories were defined in terms of truth tables. Both subsequently rejected the view that such logical rules were generally sufficient and began to examine naturalistic categories based on the idea of family resemblance. These two stages of development mostly involved exemplars in the form of static snapshots. If research on non human animal categorization continues to track the preceding development of Wittgenstein's views, the future of research on non human animal categorization will see a greater emphasis on how local temporal structure of stimuli determines membership in a category. These developmental stages are illustrated by examples from research on avian local and global levels of visual perceptual analysis, categorization of rectangles and moving objects, artificial grammar learning, and by general differences between what are labeled as molar and molecular analyses of learning and memory.

W. Todd Maddox
University of Texas

Multiple Systems of Category Learning

Categorization is a primary component of many behaviors of all organisms. Categorization allows the organism to respond differently to different classes of objects or events such as edible vs. poisonous foods, and to prey vs. predators. A growing body of research suggests that the learning of different types of category structures is mediated by different categorization and memory systems. Whereas most multiple systems theorists agree that one system relies on explicit (rule-based) reasoning and another uses some form of implicit learning, there is disagreement about the nature of the implicit system. Some argue for an exemplar- based system, some for a perceptual representation system, and others for a procedural learning-based system. In this paper I review the literature in support of multiple systems, and conclude that the evidence supports a procedural learning-based system, and thus an association with motor performance. A recently proposed neurobiological theory that assumes a procedural learning-based implicit system is outlined, and results from several studies that support the theory are reviewed.

Edmund Fantino
University of California-San Diego

Behavior-Analytic Approaches to Decision-Making

Behavior analysis has much to offer the study of phenomena in the area of judgement and decision making. We review several research areas that should continue to profit from a behavior- analytic approach. We focus on two such areas: (1) the study of base-rate neglect, a notorious reasoning fallacy and (2) the study of the sunk- cost effect, which characterizes ill-conceived investment decisions. In each case we review studies with humans and pigeons as subjects.

Richard Schuster
University of Haifa

A behavior analyst returns to cooperation learning

Probably the most widespread type of cooperation common to humans and animals occurs when individuals coordinate actions for joint outcomes, e.g., in cooperative hunting and aggression. Under free-ranging conditions, cooperative coordination is a highly social phenomenon performed by familiar individuals who use each other when coordinating, develop complementary roles and ranks, and compete over allocation of shared outcomes. A different perspective, shared by behavioral ecology, game theory and behavior analysis, explains the existence and performance of cooperation from the assumption that each participant is guided by "instrumental rationality," i.e., maximization of expected utility from individual behavior and its payoffs. Support however has come from a methodology introduced by Skinner that minimizes the influence of social factors by physically isolating anonymous subjects in separate chambers and reinforcing individual acts. Although the reinforcement contingency remains social, behavior and reinforcement become non-social events indistinguishable from those used in the study of individual learning. Our research incorporates elements of free-ranging cooperation with a laboratory model that reinforces pairs of laboratory rats Rattus norvegicus for coordinating movements within a shared chamber. Under these conditions, subjects learn to work together by means of social cues and complementary behaviors. Moreover, results of experiments on outcome competition and choice are inconsistent with "instrumental rationality." When cooperation was reinforced with intermittent presentations of single and double outcomes, competition over the allocation of the single outcomes varied widely from "sharing" to strong dominance, without any influence on the rate or level of coordinating. In the choice experiments, subjects in a T-maze chose between one chamber where rats shuttled alone and another chamber in which pairs cooperated. The cooperation option was preferred almost 3:1 even though there was no outcome advantage from choosing either chamber since outcomes and the proportions of reinforced shuttles were matched. Overall, the failure of instrumental rationality to predict these data point to as yet undiscovered sources of reinforcement that underlie cooperation as it is expressed in the real world.

Saturday Afternoon, May 24

University of Minho

Chair: Peter R. Killeen (Arizona State University)

In the first part of this tutorial, I will introduce the basic ideas of probability theory (e.g., sample spaces, events, the axioms of the theory) and then illustrate their application in a variety of practical situations some of which related to the psychology of learning. In the second part of the tutorial, I will present a few techniques that may help in solving specific problems, techniques that include counting events, using indicator variables, and constructing simple recurrence relations. Throughout the tutorial I will put the emphasis on intuition and examples, not on formalism and proof.

University of California - San Diego

Chair: John A. Nevin (University of New Hampshire)

Signal-detection theory has been around for decades, but its ability to help one think productively about a wide array of issues is not as widely appreciated as it should be. Seemingly unrelated issues are often revealed to have a common denominator when they are considered in light of detection theory, and the flaws in some otherwise intuitively appealing ideas can be fully appreciated by contrasting those ideas with a detection theory account. This tutorial will consider the utility of slightly advanced (but still simple) signal-detection techniques, like ROC analysis, to illustrate these points.

Arizona State University

Chair: Gregory Galbicka (Aventis)

Come one, come all. See the spaces between behavior. Strange beasts revealed: Hazard survival with IRTs per Op. Transubstantiate probabilities into rates. Pit Palya machines against Shull machines. See what else the dead white statisticians Gumbel, Poisson, Bernoulli and Erlang have in common. Luced [sic] expositions on distributions. Free to every guest, a whizzo spreadsheet that chops slices and dices; just insert data and you're only a click away from seeing if the tails are heavy, the probabilities constant, the values extreme. Money-back guarantee!! Come in, come in and see the show!

Wofford College

Chair: Allen Neuringer (Reed College)

Even though chaining theory has serious problems, it remains the most common explanation for behavioral sequences. Chaining theory combines several variables known to affect behavior, such as conditioned reinforcement, discriminative stimuli, and the delay-of-reinforcement gradient. Yet, as a verbal model, it cannot predict the quantity of behavior in any situation. Can current quantitative models of these individual processes (e.g. Delay Reduction Theory, Hyperbolic Value Added model, Markov chains, etc.) be applied to simple response sequences? This tutorial explores what we need to know to create a quantitative replacement for chaining theory, and what forms such a replacement might take. This tutorial will serve as an introduction to a research area with much potential for progress.

Friday Evening, May 23 06:00 - 09:00 pm