krus96: Inverse Base-rate Effect AP

krus96R Documentation

Inverse Base-rate Effect AP


In the inverse base-rate effect, participants are trained that a compound of two cues (I + PC) leads to a frequently-occurring outcome (C), while another two-cue compound (I + PR) leads to a rarely-occuring outcome (R). The key results are that, at test, participants tend to respond 'C' to cue I on its own, but 'R' to the cue compound (PC + PR). This latter response is striking because PC and PR had been perfectly predictive of diseases C and R respectively, and disease C is more common, so the optimal response to PC + PR is 'C'. Participants respond in opposition to the underlying disease base rates.




A data frame with the following columns:


Symptom presented. Take values: I, PC, PR, PC+PR, I+PC+PR, I+PCo, I+PRo, PC+PRo, I+PC+PRo, as defined by Kruschke (1996).


Response made. Takes values: C, R, Co, Ro, as defined by Kruschke (1996).


Mean probability of response, averaged across participants.


Wills et al. (n.d.) discuss the classification of these data as a Auxilliary Phenomenon, rather than a CIRP (Canonical Independently Replicated Phenomenon). In brief, these particular results have been independently replicated, but are arguably not the best exemplar of the known phenomena in this area (in particular, they lack a demonstration of the shared-cue effect in IBRE). Auxilliary Phenomena may be included in catlearn if are the subject of a simulation archived in catlearn.

The data are from Experiment 1 of Kruschke (1996), which involved the diagnosis of hyopthetical diseases (F, G, H, J) on the basis of symptoms presented as text (e.g. "ear aches, skin rash"). Participants were trained with feedback across 15 blocks of 8 trials each. They were then tested without feedback on 18 test stimuli, each presented twice.

The data are as shown in Table 2 of Kruschke (1996). The data are mean response probabilities for each stimulus in the test phase, averaged across the two presentations of the stimulus, the two copies of the abstract design, and across participants.


Andy J. Wills, René Schlegelmilch


Kruschke, J.K. (1996). Base rates in category learning. Journal of Experimental Psychology: Learning, Memory, and Cognition, 22, 3-26.


Wills et al. (n.d.). Benchmarks for category learning. Manuscript in preparation.

See Also


catlearn documentation built on March 26, 2022, 1:07 a.m.