Animal and algorithm performance in TI paradigm
Jensen Greg, Fabián Muñoz, Yelda Alkan, Vincent P. Ferrera, Terrace Herbert · Figshare · 2015
Performance on non-terminal stimulus pairs (i.e. those excluding stimuli A and G) for subjects and algorithms. Trial number is set to zero at the point of transfer from adjacent-pair-only training to testing with other pairs. (A) Smoothed response accuracy for three rhesus macaques over 200 trials of adjacent pair training, followed by the first 400 trials of responding to all pairs. Performance is divided into the adjacent pairs (BC, CD, DE, and EF) in red, the pairs with an ordinal distance of two (BD, DE, and CF) in orange, those with a distance of three (BE and DF) in green, and the pair with a distance of four (BF) in blue. Subjects show an immediate distance effect (i.e. increased accuracy as a function of ordinal distance between stimuli) from the first transfer trial. (B) Simulated performance using the betasort algorithm, using each monkey’s maximum-likelihood model parameters for each session. Since these results are simulated, lines are plotted for all distances at all times, to show how the algorithm would respond had it been presented with trials of each type. Like the monkeys, the algorithm displays an immediate distance effect. (C) Simulated performance using the betaQ algorithm, with maximum-likelihood parameters. Although a small distance effect is observed, performance remains close to chance throughout training. (D) Simulated performance using the Q/softmax algorithm. Performance remains strictly at chance throughout adjacent-pair training, and only begins to display a distance effect after the onset of the all-pairs trials. (E) Performance of human participants given 36 trials of adjacent-pair training, followed by 90 trials of non-adjacent pairs only, and finally 42 trials of all pairs. Unlike the monkeys, participants rapidly acquire the adjacent pairs, and show only a mild distance effect at transfer. (F-H) Simulations based on human performance using the three algorithms, analogous to figures B through D. As in the monkey case, Q/softmax displays no distance effect at all until non-adjacent pairs are presented. An updated version of this figure appears in Jensen et al. (2015) "Implicit value updating explains transitive inference performance: The betasort model," PLOS Computational Biology, 11, e1004523.