Circle-elipse discrimination in monkeys: Comparison of titration schedules with the method of limits

Abstract

In order to make a direct comparison, psychometric functions were derived from two titration schedules and from the method of limits. Three vervet monkeys were reinforced with food pellets for choosing a response key with a back-projected circle from among 7 keys with identical elipses. Elipse ratios from 0.400 to 0.985 were available in 18 discrete steps. The schedules were tested in this sequence: (1) a titration schedule with different up-and-down transformed response rules. Training began with a rule of 3:1 indicating that after three correct responses the next trial presented elipse ratios one step higher. A change to a lower ratio occurred after one incorrect response. Subsequently, up-and-down rules from 1:3 to 9:1 were tested, one rule per session. (2) the method of limits. One elipse ratio was tested per session first in a sequence of increasing then of decreasing ratio. (3) a titration schedule with sequential likelihood ratio tests. A target probability of correct responses was approached by deciding after each trial whether to change to a larger or to a smaller elipse ratio. Psychometric functions derived from schedules 1 and 2 coincided. Schedule 3 produced oscillations in the difficulty of the presented stimuli. The conclusion is that discrete trial titration schedules with different up-and-down transformed response rules provided reliable measures of threshold in a visual shape discrimination. Performance was more stable and the programming simpler with this schedule than with the sequential likelihood ratio test.