| Abstract: | Each of four groups of monkeys were trained on a different simultaneous discrimination procedure involving a vertical line as the correct choice. Each group, after acquiring the discrimination, was tested for generalization along the dimension of line tilt. Monkeys that learned to select the vertical line when the alternative choices were distinguished from the correct choice by two aspects (brightness and absence of line) showed almost complete tilt generalization (flat gradient). Monkeys that learned to select the line when the alternatives were distinguished only by the absence of the line showed poor tilt discrimination (generalization gradient slightly peaked at vertical). Monkeys developed a good tilt discrimination when nonvertical lines were gradually introduced by progressively darkening them on the previously blank alternatives. Monkeys developed a tilt discrimination with the lowest error rate when only horizontal alternatives were gradually introduced and then pairs of alternatives progressively closer to vertical were made available. |
