Peter J. Urcuioli - US grants

This proposal addresses the general question of how codes or labels attached to certain events in the environment influence attention to those events. The "codes" studied in the research proposed here are overt forms of instrumental behavior which are differential with respect to the events coded. The research focuses on the conditions under which attention to these coding responses compete with attention to the external events which give rise to them. More specifically, the studies address questions regarding how selective attention to a coding-response cue changes as a function of how noticeable it is (its relative salience), how well it predicts biologically important events in the environment (its relative validity), and its temporal relationship with respect to the external cue which supports it. Towards this end, pigeons will be trained on two-choice conditional discrimination tasks in which different visual stimuli (the external events) and different patterns of responding (the coding responses) provide redundant cues for correct choices. Attention to each cue is then assessed following task acquisition by presenting it separately. Overshadowing and blocking procedures will be used to manipulate the degree to which subjects attend to each cue during training. The results from these studies will have important implications for attentional mechanisms involving response-produced cues, expectancies of future reward events, and redundant stimuli whose presence depends upon the occurrence of other environmental stimuli. They will also bear upon the general issue of whether or not similar associative processes underlie instrumental and classical conditioning. The project has potential clinical relevance for psychological and psychosomatic dysfunctions for which successful therapeutic intervention depends upon the patient's ability to selectively attend to inappropriate behavioral responses.

DESCRIPTION (Adapted from applicant's abstract): The goal of this project is to understand how non-human animals come to treat physically different stimuli as behavioral equivalents in ways that cannot be explained by perceptual similarity. Using standard, two-choice conditional discriminations common to studies of other basic learning and memory processes, the project will assess the effectiveness of different training experiences in producing acquired stimulus equivalences and why some experiences are more effective than others despite considerable overlap. In the proposed studies, different discriminative stimuli will initially share a common relationship to another stimulus and/or response. Afterwards, equivalences between the different but commonly associated discriminative stimuli will be evaluated by directly training new responses to some of them and then observing whether or not the remaining stimuli also control the new responses. In short, will the new behavioral function generalize from the directly trained to the untrained (test) stimuli, as expected if the sets are equivalent? A major focus of this research is to assess whether or not conditioning of a common, mediating representation to the discriminative stimuli during training is the source of such generalization and what that mediating representation might be. The results will be important for understanding how novel behavior, one notable human characteristic, can also emerge in other animals despite the absence of direct (reinforced) training and any known language ability.

[unreadable] DESCRIPTION (provided by applicant): Acquired equivalence reflects conditioning processes underlying the ability to categorize physically dissimilar events into groups and, with it, the ability to respond appropriately to new events or new combinations of events without direct training. The goal of this proposal is to identify some of the origins of acquired equivalence classes, their effects on stimulus control, and the nature of the events that may join them. Specifically, the proposed experiments will study the role of common responses in generating classes of equivalent stimuli and serving as indices for them, and examine the potential for the responses themselves to become class members. Pigeons will be trained on a variety of two-choice conditional and simple discriminations in which multiple stimuli (including those arising from their own behavior) occasion the same subsequent response or response pattern. Afterwards, class formation will be assessed through transfer-of-control tests that have been effective in demonstrating acquired equivalence and other forms of stimulus control in previous studies of animal cognition and human categorization. Besides providing a clearer picture of how common responses promote and reveal acquired equivalence classes, the proposed work will provide data relevant to a recent hypothesis that equivalence classes in general may arise directly from reinforcement contingencies and, thus, include responses as members. The results will also provide important and needed clarifications of recently reported equivalence-like effects of potential theoretical import and will create additional points of contact and comparison between animal categorization and the behavior-analytic literature on emergent behavior. [unreadable] [unreadable]

DESCRIPTION (provided by applicant): Categorizing physically dissimilar stimuli such as objects, words, sounds, and other sensory events into the same class is fundamental to meaning, comprehension, and other aspects of cognitive development and functioning. This application studies fundamental learning processes that underlie stimulus-class formation and the resultant emergence of novel behavior, a characteristic of normal human development that is often deficient and in need of behavioral remediation in individuals with developmental disabilities. The broad objective is to demonstrate that even in the absence of language, establishing particular conditional relations between specific pairs of stimuli via reinforcement versus non-reinforcement yields sets of interchangeable stimuli (viz., stimulus classes) as evidenced by the subsequent ability to respond appropriately to novel, untrained combinations of those stimuli. The project aims are to show how such learning yields well-defined instances of emergent behavior rarely (if ever) seen in non-human animals, rigorously test the predictions of a model which assumes that ordinal position is coded as part of a stimulus' functional characteristics, and test the hypothesis that routinely non-reinforcing certain stimulus combinations while reinforcing other combinations generates stimulus classes containing the elements of the latter. In all of the proposed research, non-verbal animals possessing other, established categorization abilities (pigeons) will be concurrently trained on go/no-go matching tasks in which certain sequences of sample and comparison stimuli end in reinforcement whereas others do not. Later, stimulus-class formation will be tested by presenting novel sequences of those same stimuli. Besides revealing emergent effects, these tests will simultaneously evaluate the ordinal-specific properties of the hypothesized functional stimuli and the importance of continual exposure to non-reinforcement (as well as reinforcement) throughout training. Together, the expected pattern of findings will demonstrate that basic, general learning processes - in particular, the reinforcement contingencies used to establish conditional stimulus relations - can generate stimulus classes even in the absence of language and its neural structures and pathways. PUBLIC HEALTH RELEVANCE: Understanding that disparate things like objects, pictures, and written and spoken words can be related is fundamental to meaning, categorization, language comprehension and other characteristics of normal development and functioning. This application examines basic reinforcement and stimulus control processes that provide a foundation for such understanding and for yielding novel, generative behavior. The project underscores the possible involvement of an unrecognized or easily overlooked feature of stimuli - their temporal or ordinal properties - that may help to explain variation in equivalence training outcomes observed with normal and intellectually disabled populations. Clarifying the nature and influence of these processes is the first step toward the development of comprehensive diagnostic tests and effective treatment interventions for individuals with intellectual impairments. In addition, the discrimination paradigm used in the project has some distinctive advantages over more commonly used behavioral procedures in the developmental disabilities field, thus providing a promising alternative for establishing stimulus classes when other approaches fail.