Mary C. Potter - US grants

When people view or listen to continuous sequences of words or scenes, as we do when we look around, read, listen, or watch TV, a series of conceptual representations is activated. These are the raw material for identification and comprehension of words, pictures, and sentences, and indeed for intelligent thought more generally. A cluster of phenomena indicate that the meaning of what we see is understood very quickly, in less than 100 ms after we first look at a word or a scene. Moreover, the meaning of a sentence can be comprehended and remembered when each word appears for only 100 ms, in a rapid sequence. However, words that do not hang together are almost immediately forgotten when presented that rapidly, suggesting that understanding may be swift, but it is also fleeting. These phenomena led to the hypothesis that a new form of short-term memory, "conceptual short-term memory" or CSTM (Potter, 1993, in press), is used in the earliest stages of processing that lead to understanding. The proposed research addresses questions about CSTM: What the characteristics of this memory store? What cognitive processes can be carried out using CSTM? What is represented in CSTM-- lexical entities, phonological forms, meanings and propositions, or all of these? Is CSTM distinct from sensory memories and from conventional short-term memory and working memory? If so, how does it relate to these other forms of memory, and to long-term memory? The CSTM hypothesis raises new issues and offers new interpretations of existing work. The goal of the proposed research is to test specific claims of CSTM and to answer new questions about early processing. Because conceptual activation and structuring in CSTM occur much faster than has been supposed in many theories of working and short-term memory, special methods are required to study these processes. The proposed experiments use rapid serial visual presentation (RSVP) or words in lists and sentences because this method requires viewers to structure the information as it appears. Otherwise they will lose most of it, because unconnected words evidently cannot be encoded into STM/WM at such a high rate. The research is expected to contribute to our understanding of cognitive skills such as reading and visual search.

[unreadable] DESCRIPTION (provided by applicant): The overall aim of the proposed research is to address theoretical questions about how people perceive, comprehend, and remember words and pictures. The research focuses on the early stages of processing in conditions of overload--when stimuli appear in a rapid sequence or when two targets are in close temporal proximity or both. Overload conditions can tell us how people allocate their attention and how they consolidate memory of a stimulus. In the experiments two words or pictures are presented close in time and therefore compete for limited processing resources. Potter, Staub, & O'Connor (2002) reported a surprising shift in attention between two words presented at stimulus onset asynchronies (SOAs) varying between 13 and 213 ms (the words appeared in adjacent streams of stimuli). At very short SOAs the second word was more likely to be reported, whereas at an SOA of 213 ms the reverse was the case, and the second word was often missed (an attentional blink). A two-stage competition theory is proposed: In Stage 1 detection of a possible target draws attentional resources, but attention is labile. If a second potential target is detected very soon after the first the two targets compete for resources. When one of the targets is identified as a particular word that word enters Stage 2 for consolidation in short-term memory. Stage 2 is serial: while one target is being processed, the other must wait in Stage 1 and may be forgotten. The 15 proposed experiments will test specific predictions of this model and will extend it to new domains. A computational version of this model will be developed. The long-term goal of the proposed research is to extend conclusions from this work to perception of the environment in normal viewing conditions, in order to answer important practical as well as theoretical questions about visual attention. [unreadable] [unreadable]