Learning Abilities in Yearling Horses Using the Hebb-Williams Closed Field Maze

Abstract

The Hebb-Williams closed-field maze was used to measure learning ability of 15 yearling horses. Horses were accustomed to the apparatus and experimental procedure for 10 days. A new problem was then presented to the animals each day for 12 days, with eight trials per problem. Total excess entry scores (TE × ES) varied widely upon presentation of new problems, but regression analyses revealed no progressive day effects. TE × ES for individual horses ranged from 500 to 1,884, with a mean of 917. Rapidity-of-learning scores (%R) averaged 78, indicating that the horses made most of their errors during the first four trials of each problem. Mean visual ability scores (%P) averaged 40, suggesting that the animals were more efficient in solving problems with direct visual solutions. However, problems with visual solutions generally involved fewer physical barriers, giving horses less opportunity to make errors. Statistical analyses revealed no significant differences in any of these variables due to sex or prior learning experience. Excess entry scores from the first four trials of each problem set (FE × ES) were highly related (P<.01) to TE × ES in both simple (r=.953) and rank correlations (r=.88). However, neither of these scores was related to % R scores in simple correlations and they were negatively related to %R in rank correlations, suggesting that TE × ES and %R measured two different indices of learning.