Research into the way information is processed has indicated that it is undervalued. There are suggestions that this may be the result of central mental processes, response bias or the method of measuring subjective probability. The research described here used confidence ratings and decision times as independent indices of subjective probability. These two measures were found to correspond closely in the findings. The results show two distinct effects: one is to overestimate the probability of unlikely events (Cohen's inertial-Ψ effect) and the other is to underestimate the value of information in changing the probabilities (a ‘conservatism’ effect). However, not only is it the case that information is undervalued, but the relationship between information and subjective probability is different from the relationship between information and statistical probability. Subjective probability appears to increase linearly with the antilog of the informaton in bits. The possible meaning of this result is discussed.