Is memory for stimulus magnitude Bayesian?

Area of Science:

• Cognitive Psychology
• Memory Research
• Perceptual Learning

Background:

• Theories of memory suggest recall may involve Bayesian weighting of stimulus magnitude and category central tendency.
• Previous models proposed memory is constructed from long-run probabilities of category membership.

Purpose of the Study:

• To investigate if memory for stimulus values follows a Bayesian weighting model.
• To examine how category information and stimulus magnitude influence memory recall.

Main Methods:

• Five experiments were conducted using geometric figures from two distinct size categories.
• Participants reproduced the remembered size of presented stimuli.
• Data analysis focused on biases in reproduction toward category means and stimulus distributions.

Main Results:

• Memory reproductions were biased toward the mean of the combined stimulus distribution, not individual category means.
• The size of the stimulus from the preceding trial significantly influenced current reproductions.
• These findings challenge the notion of memory as a simple construction from category probabilities.

Conclusions:

• Memory recall for stimulus values is not solely a Bayesian weighting of stimulus magnitude and category central tendency.
• Memory construction appears to be influenced by broader distributional information and sequential trial effects.
• Existing models may need revision to account for these observed biases in memory recall.