Comparing negative patterning and biconditional discrimination in a simulated foraging task

Abstract

How humans solve patterning and biconditional discrimination is a topic of continued theoretical debate. Some theories assume that solving negative patterning of A+, B+, AB0 is contingent upon learning to associate outcomes with individual stimuli, whereas others assume the addition of a unique configuration of the compound is needed to solve it.Research in humans has found evidence to support both theories. However, a problem with human associative learning experiments is the use of arbitrary reinforcement stimuli, which may cause humans to use rule-based learning rather than associative learning. This paper reports a study of patterning and biconditional discrimination that uses a novel simulated foraging task in which all trials involve stimulus compounds to provide a clearer test of the competing theories. If given a pair of eyes, a nose, and a mouth, any normally functioning human being will see a face. But what exactly constitutes a face? And why do we no longer observe the eyes, nose, and mouth in isolation when they are presented simultaneously? Is the product of a face embedded into each individual facial feature, and is only conceptualize after a certain sum of individual facial features reaches a threshold? Or do we treat faces as unique stimuli, as something completely distinct from individual facial features? Faces can be thought of as representations from a pattern of information. Take away the pair of eyes or the mouth, and a face will likely not be readily observable. So, how does the brain create configurations, such as a face, from a pattern of otherwise isolated stimuli? The answer may lie in how humans associate outcomes with individual stimuli versus stimuli in compound.A challenge to associative learning researchers is to explain how animals solve negative patterning and biconditional discrimiantions (Pavlov, 1927;Rescorla, 1972; Saavedra, 1975;Whitlow & Wagner, 1972;McLaren & Macintosh, 2002;. Because each discrimination cannot be solved by linear properties they are particularly unique. In patterning discriminations, individual stimuli and their compound are each paired with outcomes. In positive patterning, the individual stimuli are not reinforced, whereas the compound of the same stimuli is reinforced (A0, B0, AB+). In negative patterning, the individual stimuli are reinforced, whereas the compound is not reinforced (A+, B+, AB0). In a biconditional discrimination, four stimuli are presented in compound in such a way that each individual stimulus is both 2 reinforced and non-reinforced (CD+, EF+, CE0, DF0). A stimulus is anything that represents a detectable change in the environment, which influences or causes a temporary increase of physiological activity or response. In experiments involving non-human animals, auditory and visual cues are normally used as stimuli. For example, researches designate cue A as an auditory stimulus (usually a specific tone), and cue B as a visual stimulus (a light), creating a fundamental modal distinction between each stimulus. The compound AB occurs ...

When patterning discriminations are harder than biconditional ones: A cue constellation approach.

“…3Loatman (2013) used the design described in Experiment 1 but did not include the irrelevant cue compound discrimination task. For the simulation, the Rescorla-Wagner model was used with configural cues turned off (so they could be controlled by the experimenter), and the saliences of distinctive and configural cues were either .1 versus .025 (for the relatively low salience configural cues) or .025 versus .1 (for the relatively high salience configural cues).…”