Adaptive Behavior, 6 (1)

A new neuronal model of classical conditioning is proposed. The model, called the delay-producing connections (or DPC) model, is an extension of Klopf's (1988) drive-reinforcement model and Sutton and Barto's (1981) model. The DPC model makes two contributions: It represents the trace of each conditioned stimulus (CS) by a differential equation; and it replaces each CS in the activation rule by its CS trace, which is assumed to be equal to the CS eligibility. The DPC model reproduces the usual shape of a conditioned response, the curve of efficacy of conditioning as a function of the interstimulus interval (ISI), the dependence of the optimal ISI on CS duration, the extinction of a conditioned response [even for long-lasting CSs as opposed to Klopf's (1988) model], and several other properties of classical conditioning.

Key Words

Pavlovian conditioning; connectionism; neuronal model; real-time model

Pages 63-129

Escape, Avoidance, and Imitation: A Neural Network Approach

By Nestor A. Schmajuk and B. Silvano Zanutto

Abstract

We present a real-time neural network that integrates classical and operant processes to describe how animals learn to escape and avoid an aversive stimulus either by trial and error or by imitation. It is assumed that through classical conditioning animals build an internal model of the environment and that through operant conditioning animals select from alternative responses. The internal model of the environment provides predictions of the aversive stimulus based on environmental stimuli and the animal's own responses, and these predictions are used to train the operant conditioning block to generate responses that minimize the aversive stimulus. Computer simulations show that the model correctly describes many of the features that characterize escape and avoidance.

The network also is able to describe the imitation of a demonstrator by an observer. During the demonstration, a neural network representing the observer stores classical associations between environmental stimuli and the demonstrator's responses and aversive stimuli, and these associations serve to train the operant associations during the observer's performance. It is assumed that the demonstrator's responses evoke a representation of identical responses in the observer and that the demonstrator's unconditioned response to the aversive stimulus serves as an aversive reinforcer for the observer.

The network contributes to a general theory of adaptive behavior and is relevant to the design of autonomous systems that learn either through trial and error or through imitation.

Key Words

escape; avoidance; learned helplessness; two-factor theory; neural network; classical conditioning; operant conditioning; imitation

Pages 131-161

An Autonomous Agent Navigating with a Polarized Light Compass

By Dimitrios Lambrinos, Marinus Maris, Hiroshi Kobayashi, Thomas Labhart, Rolf Pfeifer, and Rüdiger Wehner

Abstract

One of the fundamental abilities required in autonomous agents is homing. Natural agents--for instance, desert ants--solve the homing problem mainly by using path integration within an egocentric frame of reference. When employing such a mechanism, compass information for determining direction is necessary, and the precision of the compass will have a crucial effect on the precision of homing. For deriving compass information, certain insects use the pattern of polarized light in the sky that arises due to scattering of sunlight in the atmosphere (polarized light compass). The analysis of skylight polarization is mediated by specialized photoreceptors and neurons in the visual system. Inspired by the insect's polarized light compass, we have constructed a polarization compass that was employed successfully on the mobile robot Sahabot. Three models for extracting compass information from the polarization pattern of the sky were tested. In this article, we describe the navigation system and report results of experiments performed with the Sahabot in one of the natural habitats of the desert ant Cataglyphis in North Africa.

Key Words

autonomous agents; robot navigation; polarization vision; skylight compass

back to TOC, back to top

Adaptive Behavior, 6 (1)

Adaptive Behavior

Volume 6, Number 1

Summer 1997

Table of Contents

Herbert L. Roitblat

Editorial

Alain Mignault and Anthony A. J. Marley

A Real-Time Neuronal Model of Classical Conditioning