Subjective probability in choice-reaction experiments

https://doi.org/10.1016/0022-2496(69)90030-3Get rights and content

Abstract

Both choice-reaction time (CRT) and probability of error are known to depend (a) on signal probability, and (b) on the preceding sequence of signals. These two effects, (a) and (b), are shown to require a common explanation. A review of Bertelson's idea of “preparation”, which is a putative explanation of (b), is followed by a mathematical model for two-choice experiments, based on an information theory of continous response sets. The subject has an information store of fixed capacity, whose contents are updated after each trial of the experiment; moreover, the store is divided between information about signal frequency and information about the frequencies of repetitions and alternations. Expectancy (or subjective probability) is determined, in part, by the contents of the store on each trial; CRT and the probability of error are thus related to the preceding sequence of signals by an underlying probability learning process. A quantitative account is given of some previous results. Certain admissible variants of the model and its possible application to probability learning experiments are discussed.

References (22)

  • M.P. Friedman et al.

    Long-term probability learning with a random schedule of reinforcement

    Journal of Experimental Psychology

    (1968)
  • Cited by (0)

    This paper was prepared with support from a grant provided by the Science Research Council. I am grateful to Professor P. Bertelson, Dr. D. E. Broadbent and Professor M. P. Friedman for their criticism of earlier drafts of this paper.

    View full text