Sequential Assessment and Decision-Making in Humans

In two experiments inspired by theories of sequential information-gathering tasks such as sequential mate assessment, human subjects were asked to view sequences of pictures of mythical birds, each of which varied in value according to one of three stimuli, namely tail length, band width or intensity of black and carried an associated score. The subjects chose one picture from the sequence so as to maximise their points scored. Only one picture could be seen at a time, if a picture was rejected it could not be revisited and there was a small constant devaluation of each successive picture. In the first experiment two random sequences of each of the three stimuli were viewed and subjects were given no prior information on the population of pictures, nor was the devaluation specified. In the second experiment, which used only pictures which varied in tail length, subjects each saw a different random sequence and then the same fixed sequence. In addition, they were told there was either a 5%, 10% or 20% devaluation of succesive birds, and half the subjects were given prior information on the population to be viewed. Thus in this experiment scores could be corrected for position in the sequence. To summarise the major results: 1. Choice was made at an early stage in the sequence, usually the 3rd, 4th or 5th picture, and this was consistent in both experiments. 2. In Experiment 2, subjects given prior information or smaller devaluation functions chose later in the sequence. 3. In Experiment 2 there were no differences in the corrected scores achieved by subjects in our treatments of prior information or devaluation, but subjects achieved higher absolute scores as the devaluation decreased. 4. In both experiments the value of the chosen picture was positively correlated with the mean value of the prior sequence seen when absolute scores are used, but in Experiment 2 this effect disappears when corrected scores are used. 5. A large positive jump was a cue to stop in both the random and fixed sequences. Identification of relevant variables and patterns of decision making may further enlarge our understanding of sequential assessment in humans and other animal species. © E.J. Brill, Leiden, 1995