Spectral generalization gradients from pigeons as a function of response latency

Abstract

Generalization gradients on wavelength were obtained from pigeons using a discrete trials procedure. For a group tested with 3 sec trials, it was found that gradients produced by responses of shorter latencies showed greater extent of generalization than did gradients produced by responses of longer latencies. This relationship was also shown to hold when latencies were experimentally controlled by training different groups of animals on trials of differing lengths.In an earlier experiment (Holland &: Baker, 1967) generalization gradients on wavelength were obtained from pigeons using a brief discrete trials procedure and employing probability of peck as a response measure. The current paper examines the relationship between the latency of the responses and the extent of the generalization gradients produced at given latencies. Subjects Twenty-four White Carneaux pigeons were used, maintained at 80% of their ad lib weights. All were experimentally naive. ApparltusThe pigeons worked in four identical darkened cubical experimental chambers. The keys were springloaded plates of Plexiglas illuminated from behind by Bausch and Lomb Model 33-86-25 monochromators. Individual shutters for each box were solenoidactivated plates mounted in the path of the beam. Reinforcements were provided from grain magazines accessible through 3 in. circular holes below the keys. All control, timing, schedule, and analysis functions were provided by a UNC computer operating on-line. Responses were sensed by the UNC. The latency of the responses was timed by a clock internal to the LINC which operated in .05 sec units. Method Experiment 1. Following magazine training, 12 birds were trained by the method of approximation to peck a key illuminated by 550 mfJ.. After 100 continuously reinforced responses with the shutter open, 6 sec discrete trials were introduced. The animals were run daily on discrete trials for the next 13 days, during which certain parameters were slowly changed: percent reinforcement was reduced from 100% to 10%; trial time was reduced from 6 to 3 sec; and daily number of trials given was increased from 100 to 200. The animals were then run for six days on 3 sec trials and 10% reinforcement. Throughout training and testing, trials were terminated either when the allotted trial time had elapsed or when a peck occurred. During the last six days of training all birds were responding essentially to every trial.