Recognition by the pigeon of stimuli varying in two dimensions

The generalization-across-dimensions model applied to conditional temporal discrimination

Can simple choice conditional-discrimination choice be accounted for by recent quantitative models of combined stimulus and reinforcer control? In Experiment 1, two sets of five blackout durations, one using shorter intervals and one using longer intervals, conditionally signaled which subsequent choice response might provide food. In seven conditions, the distribution of blackout durations across the sets was varied. An updated version of the generalization-across-dimensions model nicely described the way that choice changed across durations. In Experiment 2, just two blackout durations acted as the conditional stimuli and the durations were varied over 10 conditions. The parameters of the model obtained in Experiment 1 failed adequately to predict choice in Experiment 2, but the model again fitted the data nicely. The failure to predict the Experiment 2 data from the Experiment 1 parameters occurred because in Experiment 1 differential control by reinforcer locations progressively decreased with blackout durations, whereas in Experiment 2 this control remained constant. These experiments extend the ability of the model to describe data from procedures based on concurrent schedules in which reinforcer ratios reverse at fixed times to those from conditional-discrimination procedures. Further research is needed to understand why control by reinforcer location differed between the two experiments.

Theory of reinforcement schedules

The three principles of reinforcement are (1) events such as incentives and reinforcers increase the activity of an organism; (2) that activity is bounded by competition from other responses; and (3) animals approach incentives and their signs, guided by their temporal and physical conditions, together called the "contingencies of reinforcement." Mathematical models of each of these principles comprised mathematical principles of reinforcement (MPR; Killeen, 1994). Over the ensuing decades, MPR was extended to new experimental contexts. This article reviews the basic theory and its extensions to satiation, warm-up, extinction, sign tracking, pausing, and sequential control in progressive-ratio and multiple schedules. In the latter cases, a single equation balancing target and competing responses governs behavioral contrast and behavioral momentum. Momentum is intrinsic in the fundamental equations, as behavior unspools more slowly from highly aroused responses conditioned by higher rates of incitement than it does from responses from leaner contexts. Habits are responses that have accrued substantial behavioral momentum. Operant responses, being predictors of reinforcement, are approached by making them: The sight and feel of a paw on a lever is approached by placing paw on lever, as attempted for any sign of reinforcement. Behavior in concurrent schedules is governed by approach to momentarily richer patches (melioration). Applications of MPR in behavioral pharmacology and delay discounting are noted.

Generalizing from the Past, Choosing the Future

Behavior in the present depends critically on experience in similar environments in the past. Such past experience may be important in controlling behavior not because it determines the strength of a behavior, but because it allows the structure of the current environment to be detected and used. We explore a prospective-control approach to understanding simple behavior. Under this approach, order in the environment allows even simple organisms to use their personal past to respond according to the likely future. The predicted future controls behavior, and past experience forms the building blocks of the predicted future. We explore how generalization affects the use of past experience to predict and respond to the future. First, we consider how generalization across various dimensions of an event determines the degree to which the structure of the environment exerts control over behavior. Next, we explore generalization from the past to the present as the method of deciding when, where, and what to do. This prospective-control approach is measurable and testable; it builds predictions from events that have already occurred, and assumes no agency. Under this prospective-control approach, generalization is fundamental to understanding both adaptive and maladaptive behavior.

Testing analogical rule transfer in pigeons (Columba livia)

Categorization is an essential cognitive process useful for transferring knowledge from previous experience to novel situations. The mechanisms by which trained categorization behavior extends to novel stimuli, especially in animals, are insufficiently understood. To understand how pigeons learn and transfer category membership, seven pigeons were trained to classify controlled, bi-dimensional stimuli in a two-alternative forced-choice task. Following either dimensional, rule-based (RB) or information integration (II) training, tests were conducted focusing on the "analogical" extension of the learned discrimination to novel regions of the stimulus space (Casale, Roeder, & Ashby, 2012). The pigeons' results mirrored those from human and non-human primates evaluated using the same analogical task structure, training and testing: the pigeons transferred their discriminative behavior to the new extended values following RB training, but not after II training. Further experiments evaluating rule-based models and association-based models suggested the pigeons use dimensions and associations to learn the task and mediate transfer to stimuli within the novel region of the parametric stimulus space.

Dimensional Shift and the Transfer of Attention

Dimensional shift was examined by generating gradients of two-dimensional stimulus generalization for intradimensional and extradimensional transfer of attention. Undergraduates discriminated between lines varying in length and orientation. Stimulus values were relevant on one dimension and irrelevant on another. For intradimensional transfer, the Phase 2 transfer task involved discriminating between new values on the dimension that was relevant in Phase 1. For extradimensional transfer, Phase 2 involved discriminating between new values on the dimension that was irrelevant in Phase 1. Extradimensional transfer was learned in twice the number of trials that were required for intradimensional transfer. In Experiment 1, generalization gradients obtained at different stages of Phase 2 training showed that control by the relevant dimension was maintained throughout the intradimensional transfer. In the extradimensional transfer, however, control by the previously relevant dimension was gradually lost before control by the new relevant dimension was acquired. Experiment 2 showed that the advantage of intradimensional over extradimensional transfer could not be attributed to cue-specific stimulus generalization. Experiment 3 showed that in extradimensional transfer the irrelevant dimension in Phase 1 retarded acquisition of control by the new relevant dimension in Phase 2. Experiment 4 showed that the irrelevant dimension masked control by the relevant dimension in the generalization test, but verified the conclusion from Experiment 3 that learned irrelevance contributed to the intradimensional-extra-dimensional transfer difference.

Divided stimulus control: a replication and a quantitative model

Four pigeons were trained on a conditional discrimination. The conditional stimuli were compounds of pairs of stimuli from two different dimensions, fast versus slow cycles of red or green stimuli, and short- versus long-duration presentations of these cycles. Across conditions, the probability of reinforcers for correctly responding to each dimension was varied from 0 to 1. Discriminability, measured by log d, for stimuli on a dimension increased as the relative frequency of reinforcers for that dimension increased, replicating the results of Shahan and Podlesnik (2006). Two further conditions showed that discriminability between stimuli on each dimension was unaffected by whether the stimuli on the other dimension varied or were constant. Finally, maximal discriminability was unchanged in a redundant-relevant cues condition in which either of the stimuli comprising a compound signaled the same correct response. Davison and Nevin's (1999) model provided an excellent quantitative account of the effect of relative reinforcer frequency on discriminability, and thus of the way in which divided stimulus control is itself controlled by relative reinforcement.

Integrality/separability of stimulus dimensions and multidimensional generalization in pigeons

The authors present a quantitative framework for interpreting the results of multidimensional stimulus generalization experiments in animals using concepts derived from the geometrical approach to human cognition. The authors apply the model to the analysis of stimulus generalization data obtained from pigeons trained with different sets of stimuli varying along two orthogonal dimensions. Separable pigeons were trained with stimuli varying along the dimensions of circle size and line tilt, dimensions found to be separable in previous human research; integral pigeons were trained with stimuli varying along two dimensions of rotation in depth, dimensions that are intuitively integral and which hold special interest for theories of object recognition. The model accurately described the stimulus generalization data, with best fits to the City-Block metric for separable pigeons and to the euclidean metric for integral pigeons.

Complex learning and information processing by pigeons: a critical analysis

THREE MODELS OF CONDITIONAL DISCRIMINATION LEARNING BY PIGEONS ARE DESCRIBED: stimulus configuration learning, the multiple-rule model, and concept learning. A review of the literature reveals that true concept learning is not characteristic of the behavior of pigeons in matching-to-sample, oddity-from-sample, or symbolic matching studies. Instead, pigeons learn a set of sample-specific S(D) rules. Transfer of the discrimination to novel stimuli, at least along the hue dimension, is predicted by a "coding hypothesis", which holds that pigeons make a unique, but usually unobserved response, R(1), to each sample, and that the comparison stimulus chosen depends on which R(1) was emitted in the presence of the sample. Convincing evidence is found that pigeons do code sample hues, but there is little evidence that allows one to infer that the "coding event" must have behavioral properties. Parameters of the conditional discrimination paradigm are identified, and it is shown that by appropriate parametric manipulation, a variety of analogous tasks may be generated for both human and animal subjects. The tasks make possible the comparative study of complex learning, attention, memory, and information processing, with the added advantage that behavior processes may be compared systematically across tasks.

Interresponse-time analysis of stimulus control in multiple schedules

Interresponse-time distributions were recorded in two components of multiple variable-interval schedules that were varied over several conditions. Values of the exponent for power functions relating ratios of interresponse times emitted per opportunity to ratios of reinforcers obtained in the two components varied with interresponse-time class interval. The exponent (sensitivity to reinforcement) afforded a measure of stimulus control exerted by the discriminative stimuli. Exponents were near zero for short interresponse times, consistent with previous conclusions that responses following short interresponse times are controlled by response-produced or proprioceptive stimuli. Values of exponents increased with longer interresponse times, indicating strong control by exteroceptive stimuli over responses following interresponse times of approximately one second or longer.

Steady state responding based upon simple and compound stimuli

Two pigeons were trained to perform discrimination tasks along two dimensions, wavelength of a circular spot of light and orientation of a white line. Discriminability among stimuli along these dimensions was established for both subjects by means of a steady state testing procedure. The two dimensions were then combined by superimposing the white line upon the colored background. Subjects were given a series of tests in which a correct response could be made on the basis of either of the two components of the stimulus compounds. Discriminability among these redundant compound stimuli was found to be better than that among wavelength and tilt stimuli alone. A second series of tests was administered using both redundant and conflicting compound stimuli. The results of this test series are consistent with a response strategy in which subjects first examine both elements of a compound and then emit a choice response on the basis of the element that best predicts reinforcement.

The legacy of Guttman and Kalish (1956): Twenty-five years of research on stimulus generalization

This paper is a selective review of the methods, problems, and findings in the area of operant stimulus generalization over the 25 years since the publication of the original paper by Guttman and Kalish (1956) on discriminability and spectral generalization in the pigeon. The paper falls into five main sections, which encompass the main themes and problems stemming from the Guttman and Kalish work and its immediate successors. The first section addresses the relationship between stimulus generalization and stimulus control, as well as the variety of testing procedures and dependent variables used to measure generalization. The next section reviews the limited literature on the effects of early rearing on the generalization gradient. The relationship between discriminability among test stimuli and the slope of the spectral gradient is discussed in the third section, with emphasis upon recent reassessments of the pigeon's hue discriminability function. The fourth section reviews the topic of inhibitory stimulus control, one which developed with the discovery of the peak shift following intradimensional discrimination training. Problems of definition and measurement are discussed in conjunction with the gradient forms used to index inhibitory control. The last section is devoted to attentional effects and the two principal theories postulated to account for them. A survey of different attentional paradigms is provided and the possible role of constant irrelevant stimuli as a source of control is examined. A brief conclusion summarizes the contribution of the generalization technique toward an understanding of the nature and acquisition of stimulus control.

Reaction times of pigeons on a wavelength discrimination task

After extensive pretraining, three pigeons were exposed in 2-second trials to a random series of 14 light wavelengths, ranging in one nanometer (nm) steps from 575 nanometers to 589 nanometers. Responses to one of the wavelengths, 582 nanometers, were intermittently reinforced. The relative frequency of response approached 1.0 at 582 nanometers, and decreased with progressively higher and lower wavelengths. Reaction times shorter than about 0.2 second occurred with a low frequency that was largely independent of wavelength. Wavelength controlled the frequency of longer reaction times, but did not affect the distribution of these reaction times. Consequently, receiver-operating characteristic curves constructed by using reaction time as a rating measure did not conform to the signal-detection model, in contrast to such conformity when response rate is used in a similar way. The data suggest that stimulus onset as such triggers early response emission with some small probability; the probability of responses with longer latency is controlled by wavelength, but their time of emission is controlled by some independent process.

Methods and theories in the experimental analysis of behavior

We owe most scientific knowledge to methods of inquiry that are never formally analyzed. The analysis of behavior does not call for hypothetico-deductive methods. Statistics, taught in lieu of scientific method, is incompatible with major features of much laboratory research. Squeezing significance out of ambiguous data discourages the more promising step of scrapping the experiment and starting again. As a consequence, psychologists have taken flight from the laboratory. They have fled to Real People and the human interest of “real life,” to Mathematical Models and the elegance of symbolic treatments, to the Inner Man and the explanatory preoccupation with inferred internal mechanisms, and to Laymanship and its appeal to “common sense.” An experimental analysis provides an alternative to these divertissements.

The “theories” to which objection is raised here are not the basic assumptions essential to any scientific activity or statements that are not yet facts, but rather explanations which appeal to events taking place somewhere else, at some other level of observation, described in different terms, and measured, if at all, in different dimensions. Three types of learning theories satisfy this definition: physiological theories attempting to reduce behavior to events in the nervous system; mentalistic theories appealing to inferred inner events; and theories of the Conceptual Nervous System offered as explanatory models of behavior. It would be foolhardy to deny the achievements of such theories in the history of science. The question of whether they are necessary, however, has other implications.

Experimental material in three areas illustrates the function of theory more concretely. Alternatives to behavior ratios, excitatory potentials, and so on demonstrate the utility of rate or probability of response as the basic datum in learning. Functional relations between behavior and environmental variables provide an account of why learning occurs. Activities such as preferring, choosing, discriminating, and matching can be dealt with solely in terms of behavior, without referring to processes in another dimensional system. The experiments are not offered as demonstrating that theories are not necessary but to suggest an alternative. Theory is possible in another sense. Beyond the collection of uniform relationships lies the need for a formal representation of the data reduced to a minimal number of terms. A theoretical construction may yield greater generality than any assemblage of facts; such a construction will not refer to another dimensional system.

Reaction times in discriminations of varying difficulty: Decision modulated by arousal

Pigeons discriminated the hue of a spot of light that appeared in discrete trials. A green spot always signaled food ("green S+") and so did a red spot of constant hue ("red S+"), but on most trials a different red hue appeared and no food was given for pecks ("red S-"). The hue of red S- stayed the same during blocks of up to twelve sessions, but it changed from block to block. During a final group of sessions red S- was omitted and the percent reinforcement to the two S+ stimuli was varied. Major findings were that (1) percent response (Pct(R)) to S- varied with S+/S- similarity, describing a typical ogival psychometric function; (2) Reaction times (RTs) to both red and green S+s were minimal when the red discrimination was impossible (that is, when red S+ and S- were identical); (3) RTs to red S+ were greatest during discriminations of intermediate difficulty; (4) as Pct(R) declined during the learning of a difficult discrimination, RTs increased to red S+ as well as red S-. Most aspects of the data were reproduced by simulations with a quantitative model that incorporated reinforcement-based decision and arousal processes.

Why are artificial polymorphous concepts so hard for birds to learn?

Stimulus sets defined in terms of artificial polymorphous concepts have frequently been used in experiments to investigate the mechanisms of discrimination of natural concepts, both in humans and in other animals. However, such stimulus sets are frequently difficult for either animals or humans to discriminate. Properties of artificial polymorphous stimulus sets that might explain this difficulty include the complexity of the individual stimuli, the unreliable reinforcement of individual positive features, attentional load, difficulties in discriminating some stimulus dimensions, memory load, and a lack of the correlation between features that characterizes natural concepts. An experiment using chickens as subjects and complex artificial visual stimulus sets investigated these hypotheses by training the birds in discriminations that were not polymorphous but did have some of the properties listed above. Discriminations that involved unreliable reinforcement or high attentional load were found to approach the difficulty of polymorphous concept discriminations, and these two factors together were sufficient to account for the entire difficulty. The usual kind of artificial polymorphous concept may not be a good model for natural concepts as they are perceived and discriminated by birds. A RULEX account of natural concept learning may be preferable.

Do chickens (Gallus gallus f. domestica) decompose visual figures?

To investigate whether learning to discriminate between visual compound stimuli depends on decomposing them into constituting features, hens were first trained to discriminate four features (red, green, horizontal, vertical) from two dimensions (colour, line orientation). After acquisition, hens were trained with compound stimuli made up from these dimensions in two ways: a separable (line on a coloured background) stimulus and an integral one (coloured line). This compound training included a reversal of reinforcement of only one of the two dimensions (half-reversal). After having achieved the compound stimulus discrimination, a second dimensional training identical to the first was performed. Finally, in the second compound training the other dimension was reversed. Two major results were found: (1) an interaction between the dimension reversed and the type of compound stimulus: in compound training with colour reversal, separable compound stimuli were discriminated worse than integral compounds and vice versa in compound training with line orientation reversed. (2) Performance in the second compound training was worse than in the first one. The first result points to a similar mode of processing for separable and integral compounds, whereas the second result shows that the whole stimulus is psychologically superior to its constituting features. Experiment 2 repeated experiment 1 using line orientation stimuli of reversed line and background brightness. Nevertheless, the results were similar to experiment 1. Results are discussed in the framework of a configural exemplar theory of discrimination that assumes the representation of the whole stimulus situation combined with transfer based on a measure of overall similarity.

Anticipating by Pigeons Depends on Local Statistical Information in a Serial Response Time Task

Pigeons responded in a serial response time task patterned after that of M. J. Nissen and P. Bullemer (1987) with humans. Experiment 1 produced global facilitation: Response times in repeating lists of locations were faster than when locations were random. Response time to a spatial location was also a function of both that location's 1st- and 2nd-order local predictability, in rough agreement with the Hick-Hyman law, according to which response time is a linear function of amount of information. Experiment 2 showed that both local and global facilitation is limited to moderate response-to-stimulus intervals of about 0.50 to 2.00 s. Experiment 3 showed that response time did not depend on global statistical information. Overall, local and global performances depended on local statistical information, but global performance did not depend on global information. Local facilitation was interpreted in plain English as anticipating.

Categorizing a moving target in terms of its speed, direction, or both

Pigeons categorized a moving target in terms of its speed and direction in an adaptation of the randomization procedure used to study human categorization behavior (Ashby & Maddox, 1998). The target moved according to vectors that were sampled with equal probabilities from two slightly overlapping bivariate normal distributions with the dimensions of speed and direction. On the average, pigeons categorized optimally in that they attended to either speed or direction alone, or divided attention between them, as was required by different reinforcement contingencies. Decision bounds were estimated for individual pigeons for each attentional task. Average slopes and y intercepts of these individually estimated decision bounds closely approximated the corresponding values for optimal decision bounds. There is therefore at least one task in which pigeons, on the average, display flexibility and quantitative precision in allocating attention to speed and direction when they categorize moving targets.

Some contributions of signal detection theory to the analysis of stimulus control in animals

This paper reviews some applications of Signal Detection Theory (SDT) to the quantitative analysis of non-human animal discrimination. The basic detection model is briefly outlined and the separation of sensitivity and bias is illustrated. Several other applications and ideas are reviewed, including the rating method, some implications of signal and criterion variance, and the measurement of 'guesses' not encompassed by standard SDT. The conceptual framework and analytic tools of SDT help to clarify processes underlying stimulus control and provide direction to more complete process models.

Selective attention in animal discrimination learning

The traditional approach to the study of selective attention in animal discrimination learning has been to ask if animals are capable of the central selective processing of stimuli, such that certain aspects of the discriminative stimuli are partially or wholly ignored while their relationships to each other, or other relevant stimuli, are processed. A notable characteristic of this research has been that procedures involve the acquisition of discriminations, and the issue of concern is whether learning is selectively determined by the stimulus dimension defined by the discriminative stimuli. Although there is support for this kind of selective attention, in many cases, simpler nonattentional accounts are sufficient to explain the results. An alternative approach involves procedures more similar to those used in human information-processing research. When selective attention is studied in humans, it generally involves the steady state performance of tasks for which there is limited time allowed for stimulus input and a relatively large amount of relevant information to be processed; thus, attention must be selective or divided. When this approach is applied to animals and alternative accounts have been ruled out, stronger evidence for selective or divided attention in animals has been found. Similar processes are thought to be involved when animals search more natural environments for targets. Finally, an attempt is made to distinguish these top-down attentional processes from more automatic preattentional processes that have been studied in humans and other animals.

Receiver psychology and the evolution of multicomponent signals

Many animals produce and respond to signals made up of multiple components. For example, many avian sexual displays are highly extravagant combinations of visual and acoustic elements, and are described as being 'multicomponent'. One possible reason for the evolution of such complex signals is that they provide more reliable information for receivers. However, receivers also influence signal evolution in another important way, by how they perceive and process signals: signallers will be selected to produce signals that are more easily received. The potential role of receiver psychology in the evolution of multicomponent signals has not previously been considered; in this review I present psychological results that support the notion that two components are better received than one alone. Detection can be improved by producing two components together, thus reducing the reaction time, increasing the probability of detection and lowering the intensity at which detection occurs. Discriminability of multicomponent stimuli is also made easier through better recognition, faster discrimination learning and multidimensional generalization. In addition, multicomponent stimuli also improve associative learning. I show that multicomponency does indeed improve signal reception in receivers, although the benefits of producing components in two sensory modalities (bimodal multicomponent signals) may be larger and more robust than producing them in just one (unimodal multicomponent signals). This highlights the need for consideration of receiver psychology in the evolution of multicomponent signals, and suggests that where signal components do not appear to be informative, they may instead be performing an important psychological function. Copyright 1999 The Association for the Study of Animal Behaviour.

Mechanics of the animate

Behavior is treated as basic physics. Dimensions are identified and their transformations from physical specification to axes in behavioral space are suggested. Responses are treated as action patterns arrayed along a continuum of activation energy. Behavior is seen as movement along a trajectory through this behavior space. Incentives or reinforcers are attractors in behavior space, at the centers of basins of lowered potential. Trajectories impinging on such basins may be captured; repeated capture will warp the trajectory toward a geodesic, a process called conditioning. Conditioning is enhanced by contiguity, the proximity between the measured behavior and the incentive at the end of the trajectory, and by contingency, the depth of the trajectory below the average level of the potential energy landscape. Motivation is seen as the potential of an organism for motion under the forces impinging on it. Degree of motivation is characterized by the depth of the potential field, with low motivation corresponding to a flat field and a flat gradient of activation energy. Drives are the forces of incentives propagated through behavior space. Different laws for the attenuation of drive with behavioral distance are discussed, as is the dynamics of action. The basic postulate of behavior mechanics is incentive-tracking in behavior space, the energy for which is provided by decreases in potential. The relation of temporal gradients to response differentiation and temporal discrimination is analyzed. Various two-body problems are sketched to illustrate the application of these ideas to association, choice, scalar timing, self-control, and freedom.

Conjoint control of performance in conditional discriminations by successive and simultaneous stimuli

In a conditional discrimination, reinforcement of pigeons' responses to pairs of simultaneously presented wavelength stimuli depended on the orientation of white lines superimposed on the wavelengths. Over different conditions in Experiment 1, three wavelength differences were combined with two differences between successively presented line orientations. Measures of stimulus discriminability increased with increases in the difference between both orientation and wavelength stimuli. Conditional-discrimination performance was thus conjointly determined by stimulus disparity in the successive and simultaneous discriminations. In Experiment 2, ratios of rates of reinforcement contingent upon the two categories of correct responses were varied over several conditions for difficult and easy discriminations. Ratios of responses to wavelength pairs were sensitive to variations in the reinforcement ratio to a greater extent for the more difficult orientation discrimination than for the easier orientation discrimination. Performance in the conditional discrimination was therefore determined by the interacting effects of stimulus disparity and the relative rates of reinforcement contingent upon the two correct choices. It was concluded that the effect of temporally distant reinforcement on behavior in a prevailing schedule component is attenuated to an extent that depends on similarity of stimuli that delineate the successive components.