Accumulation of Visual Memory for Natural Scenes: A Medium-Term Memory? David Melcher (dmelcher@brookes.ac.uk) Department of Psychology, Oxford Brookes University Gipsy Lane, Oxford OX3 0BP, United Kingdom Abstract We studied visual memory for objects in natural scenes. Participants viewed photographs or pictures for periods of 1 to 20 seconds, and were then asked questions about the color, location, and identity of objects that they had seen, as well as given recognition tests. Performance improved as a function of display duration for all question types and for objects of both central and peripheral interest. On some trials, previously viewed stimuli, which had been shown 4 to 6 trials earlier with no subsequent memory test, were repeated to see if performance continued to improve across separate presentations. There was no loss of information across retests, such that memory for a display shown for 10 and then 5 seconds was equal to performance after a single trial of 15 seconds. Overall, memory performance exceeded the capacity and duration limits of short-term or working memory, supporting the idea of a medium-term visual scene memory. Introduction The ability to remember the contents of visual displays is critical for a variety of real-life tasks. Traditionally, this process has been characterized as “storage” of a few items in visual short-term memory (VSTM) followed by either information loss or transfer into long term memory (LTM). Visual short-term memory is generally considered to contain only about four items (Cowan, 2001; Vogel & Machizawa, 2004), although the exact capacity remains a matter of debate (Brockmole et al., 2002; Rensink, 2000). Unfortunately, such a limited short-term memory system would make learning about the world extremely difficult. For example, imagine that the typical view of a cluttered office desk contains at least three to four objects. It would follow that if someone looks at that desk, then it would cause the person to forget everything they had just seen unless it had already been stored in long-term memory. The dichotomy between short and long-term memory implies that people either spend a great deal of their waking life rehearsing information to keep it from fading away, or that they must be constantly filling their long-term memory with information that may later turn out to be useless. Real-life memory, as opposed to memory for relatively meaningless stimuli (letters, numbers, colored shapes), may not fit neatly into the storage model of memory. Objects in naturalistic scenes, for example, are linked with the scene context in memory rather than stored as a discrete set of objects (Melcher, 2001; Melcher & Kowler, 2001). Moreover, visual memory for objects and other details in photographs exceeds the three to four item memory limit after just a few seconds of viewing (Melcher, 2001; Melcher & Kowler, 2001; Tatler et al., 2003). The build-up and persistence of memory does not, however, depend on transfer to long-term memory, since memory for scenes did not persist across testing on separate days. The finding that memory for objects in a room exceeds both the capacity and duration of traditional short-term visual memory, while not necessarily finding a permanent place in long-term memory, suggests the existence of a “medium- term” working memory for visual scenes (Melcher, 2001; Pierrot-Deseilligny et al., 2002). There is considerable support for the idea that useful visual information accumulates across separate glances. For example, Tatler et al (2003) found that performance in answering questions about the visual details of a photograph increased over a 10 second interval. Henderson and colleagues have reported accurate recognition memory for the visual features of objects shown several minutes earlier (Henderson et al., 2003). In addition, there is evidence that traditional measures of short-term memory underestimate the visual processing of a larger number of items in the brain (VanRullen & Koch, 2003) The goal of the present study was to examine the build-up and persistence of working memory for different visual attributes in a complex image. Natural scenes contain a myriad of visual information, including color, texture, shape, absolute and relative object location, object identity, overall spatial layout and scene gist. Recent studies suggest that different visual features may have different rates of memory accumulation and decay (Melcher & Morrone, 2003; Tatler et al., 2003). Such differences in memory for particular visual features might, in fact, help to explain the conflicting reports in the visual memory literature. A second, and related, issue is the role of scene-related salience in the guidance of attention and subsequent memory for objects in a scene. In studies of change detection, alterations to the scene are noticed more quickly when they occur to an item of “central interest”, rather than one that is peripheral to the gist or general meaning of the picture (Rensink, 2000). This raises the question of which types of items benefit from extended viewing of a scene. One possibility might be that salient items would show more improvement because they are viewed earlier and more often, increasing the chance that their details would be in memory, while peripheral items would tend to be ignored On the other hand, peripheral items might show the greatest benefit from longer presentations, compensating for a lack of earlier attention, while central items would lose their initial dominance in memory. 1485