Generalisation: mechanistic and functional explanations

Abstract

An overview of mechanistic and functional accounts of stimulus generalisation is given. Mechanistic accounts rely on the process of spreading activation across units representing stimuli. Different models implement the spread in different ways, ranging from diffusion to connectionist networks. A functional account proposed by Shepard analyses the probabilistic structure of the world for invariants. A universal law based on one such invariant claims that under a suitable scaling of the stimulus dimension, generalisation gradients should be approximately exponential in shape. Data from both vertebrates and invertebrates so far uphold Shepard's law. Some data on spatial generalisation in honeybees are presented to illustrate how Shepard's law can be used to determine the metric for combining discrepancies in different stimulus dimensions. The phenomenon of peak shift is discussed. Comments on mechanistic and functional approaches to generalisation are given.

Flowers help bees cope with uncertainty: signal detection and the function of floral complexity

“…430nm) that is biased in a direction away from the S-(470nm, the color that was unrewarding or punishing during training). These subjects may also show a weaker form of perceptual shift known as 'area shift' (Cheng et al, 1997;Cheng, 2002;Lynn et al, 2005), wherein responses to test stimuli are asymmetrically distributed around the S+ (e.g. S+ is 450nm, S-is 470nm, and subjects show a stronger response to test stimuli between 400 and 449nm than to stimuli between 451 and 500nm).…”

Learning-related shifts in generalization gradients for complex sounds

“…In a classic example, pigeons trained to respond to a key illuminated with a 580 nm light also responded to other wavelengths when reinforcement for responses was absent; the number of responses to newly presented wavelengths lessened with decreasing similarity to the conditioned stimulus (Guttman & Kalish, 1956). This phenomenon is seen in animals ranging from invertebrates (Cheng, 1999, 2000, 2002) to humans (reviewed by Thomas, 1993) with various testing strategies and dimensions (Shepard, 1987). Because of its ubiquity, generalization was described by Pavlov (1927) as being a fundamental associative process and by Shepard (1987) as psychology’s first law.…”

Generalization of Color by Chickens: Experimental Observations and a Bayesian Model

“…Whereas discrimination thresholds and receptor noise are fairly easy to define, it is more difficult to give an optimal model to test hypotheses about processes such as generalization and categorization. Bayesian models are increasingly used in psychology and neuroscience to develop testable predictions of optimal performance (Rao et al 2002;Tenenbaum and Griffiths 2001;Cheng 2002;Cheng and Spetch 2002). Bayesian models have also been applied directly to color perception (Brainard and Freeman 1997), but this work has concerned how representations robust to illumination changes can be optimally derived rather than how we generalize and categorize such stimuli.…”

“…A principal aim of our study is to consider the possibility that peak shift in birds is a consequence of generalization behavior following principles of optimal inference. Others have suggested that peak shift may be optimized according to criteria proposed by decision theory (Thomas et al 1991;Lynn et al 2005) or Bayesian inference (Cheng 2002). As these papers point out, if perception is indeed optimized for dealing with an uncertain world, it is inappropriate to make any inferences about neural mechanisms.…”