CAREER: Novel approaches to integrating color perception and color memory

Visual perception gives us information about objects in the world. For that information to be useful, however, it must be combined with memory. For example, the yellow color of a banana indicates ripeness only if it can be compared to a memory that distinguishes the yellow of ripe bananas from the green of unripe bananas. Everyday experience suggests that although sometimes color memory is very good (e.g. recalling the color of stop signs and bananas), it can also be very poor (e.g. picking out paint to match the walls at home). This everyday experience of color memory reflects a deep division between cognitive and perceptual scientists about the fidelity of color memory. How can these differences be reconciled? NSF-funded research conducted by Dr. Sarah Allred at Rutgers University aims to provide a single framework that unifies color perception and the disparate views of color memory. Human observers will make perceptual and memory judgments about the same colorful stimuli in progressively more complex environments. These data will then drive the development of a computational model that aims to describe color memory as a combination of color perception and independent noise. This view is in contrast to a prevailing account of cultural or linguistic relativity in color memory.

The relationship between perception and memory is important to fields as varied as forensic psychology (e.g. the reliability of eyewitness testimony), military reconnaissance, cognitive psychology (e.g. linguistic effects on memory) and medical diagnosis (e.g. does a dark cloud on an MRI image match memory templates of tumors?). The computational approach here represents an important advance to the study of perception and memory, because it will provide a baseline comparison for future research. This study will allow researchers to quantify how much of what is remembered is predicted by what was perceived. If memory is a predictable function of perception, there are two important implications: first, basic perceptual information will allow more sensible bounds to be placed on the reliability of memory; second, it suggests the possibility that querying memory in the appropriate perceptual environment will increase the reliability of memory. Although color is a relatively simple stimulus with which to probe the relationship between perception and memory, it is a good first approximation for more complicated stimuli like those involved in eyewitness testimony or medical diagnosis. This is because color is simple enough that its physical dimensions are well-characterized, thus making color amenable to computational modeling; in addition, unlike other simple physical stimuli such as oriented gratings, color is linked to higher order variables like emotion and language. Thus color provides a good bridge between simple physical stimuli and more complex stimuli with real-world applicability.