Low- and high-level mechanisms of perception and memory PDF Print E-mail

Studies of long-term memory for pictures of objects and natural scenes show that humans have a truly remarkable memory for meaningful visual images, correctly recognizing hundreds (even thousands) of individual pictures briefly viewed only once, a week prior to the memory test. What is the basis of this memory performance? Psychophysical studies, applying the concepts and methods of spatial vision to the study of visual memory, indicate the existence of a special low-level perceptual memory mechanism responsible for storing information about elementary features or dimensions of visual images (Magnussen, 2000).

 

This memory mechanism is closely tied to mechanisms of visual discrimination, it is responsible for the storage of elementary visual attributes in memory up to at least 24 hours, and might assist the formation and consolidation of the long-term memory of meaningful visual images. The present project investigates the relationship between the low-level storage of elementary features and the memory for complex and meaningful visual patterns.
Ongoing experiments follow two main lines of inquiry: First, an operational definition of memory is that memory starts when a stimulus is turned off. However, experiments on perceptual memory where choice reaction times are measured concurrently with the accuracy of the memory report in delayed discrimination tasks, indicate that in terms of brain processing the transition from perception to memory does not correspond to the termination of a stimulus, but occurs after about three seconds (Magnussen, 2000). Before that time, stored information is as accessible as online information. We are mapping the transition from a perceptual representation to a memory representation in greater detail, and use the results to guide us in designing fMRI experiments. The delayed discrimination technique has the advantage that the only difference between perception and memory tasks is in time interval between the stimuli to be compared. Thus it is an ideal design for fMRI experiments.

In further experiments we look at the relationship between low-level and high-level perceptual memory by comparing patterns of activation with abstract stimuli such as sinusoidal gratings with the activations produced by identifiable objects. In a recent study, Vogt & Magnussen (2007) modified classical studies of long-term pictorial memory by narrowing the range of depicted objects and scenes down to a single motif: Doors. The results showed very good memory for large samples (400) of studied pictures with little decay across two weeks intervals. By comparing simple and complex visual patterns, we may cast more light on the idea of specialized low-level memory mechanisms for elementary features of visual stimuli.

A second line of research on the relationship between low-level and high-level memory processes in visual memory relates to the proposed specialization of neighbouring areas of the temporal lobe for visual categories such as faces versus objects, and to the influence of verbal or categorical processes in pictorial memory. In a series of cognitive and fMRI experiments we include pictures of a single category of objects – doors - and pictures of faces, and compare brain activations in simple perceptual discrimination tasks, in memory tasks where a sub-set of the pictures is over-learned, and in tasks where they are coupled to biographical information.

Principal researchers CSHC
Svein Magnussen, Tor Endestad, Maria Korsnes

Collaborators
Mark W. Greenlee, University of Regensburg, Germany; Pauline Due Tønnesen & Atle Bjørnerud, Department of Radiology, Rikshospitalet, Norway