Fig. 2.

Mean [± 1 SD] psychometric functions for simulated observers. The shape of the function was logistic. The probability of responding correctly to a stimulus of magnitude x was therefore: $P(\text{Correct})=G+(1-L-G)(1/[1+e^{-\frac{x-T}{S}}])$. The Guess Rate, G, was fixed at 50%. The other three parameters—threshold (T), slope (S), and lapse rate (L)—varied randomly between observers, according to either a truncated Gaussian distribution (compliant observers) or a uniform distribution (non-compliant observers). See table for exact values for the two distributions. For more information on psychometric functions and the four parameters employed here, see Klein (2001)