Nelson Cowan - Memory and Attention in Human Cognition    

Research Specializations

  • Short-term or working memory
  • Childhood development of short-term or working memory
  • Relations between working memory and selective attention in information processing

For further details, see publications or C.V.

Some Professional Activities


Public Issues


Statement of Research Interests and Orientation

My research always has been driven by basic philosophical questions about the human mind, concerned with the most basic elements of conscious experience. What mechanisms allow human beings to experience the world as they do? Experiments on memory, attention, perception, and cognitive development address this question.

Current research focus: selective attention and working memory. "Working memory" (or short-term memory) can be defined as the small amount of information that can be kept in an accessible state in order to be used in ongoing mental tasks. For example, to comprehend ongoing language, one must hold in mind what has been said so far. In a 1988 article in Psychological Bulletin I put forward my view that what has been called working memory actually refers to two processes, a source of some confusion. Working memory refers to the automatic, temporary persistence of sensory and semantic information recently activated in the brain, and also to the inclusion of a subset of the activated information in the focus of attention. Donald Broadbent, a pioneer in the field, asked me to write a book expanding this view (1995, Oxford University Press, Attention and memory: An integrated framework). In addition to discussing memory, I offered hypotheses of how information enters the focus of attention, and of the relation between memory activation and attention.

Related to this theoretical view, I initiated a line of research on the role of attention in perception that has proved to be very informative. In 1990, Cowan et al. (Journal of Experimental Psychology: Learning, Memory, and Cognition) showed that the identification of vowels was about 20% higher (a very large effect) when the vowels were attended during their presentation than when they were ignored until just after their presentation. In a subsequent series of studies, we determined why one can notice unusual events in an ignored acoustic channel, akin to when you are engrossed in a conversation but you hear your own name within another conversation in the room. We found that this takes place only with disruption of the ongoing task; you cannot notice a strange event without impairing your ability to keep on doing what you were doing. Still, it was unclear why only some people noticed their names in an unattended channel. Conway, Cowan, and Bunting (2001, Psychonomic Bulletin & Review) showed that, interestingly, people with low working memory span are much more likely to notice their names. Their low working memory seems related to an inability to focus their attention strongly on the assigned task, so their attention is more free to wander during the task. This finding helps explain individual differences in working-memory capacity, which in turn correlates highly with intelligence and scholastic-aptitude tests.

Until recently, it has been difficult to measure the limits of working memory. In a seminal work, Miller (1956) showed that people could remember about 7 items in serial-order memory tasks. However, it was not clear what the fundamental limit was because the mental units were said to be groupings of items, or chunks. A 7-digit phone number might be recalled as a smaller number of easily-memorized chunks, e.g., 246-89-21. Unless the chunking processes can be revealed, there is no way to estimate how many chunks can be kept in working memory. In a 2001 theoretical review in Behavioral and Brain Sciences, I focused on many situations in which people presumably could not carry out chunking (e.g., because attention was diverted away from the stimuli until after they were presented). Under such circumstances, adults were able to recall 3-4 items. Research with children (Cowan et al., Child Development, 1999) showed that it was more like 2-3 items for children in the early elementary school years. Thus, this is a basic aspect of working memory that changes with development. The results also showed that being allowed to rehearse the lists improved everyone's recall but still did not reduce the difference between younger and older children. Our ongoing adult research shows that the limit of 3 to 4 chunks holds true even when multi-item chunks are memorized. Measures of this capacity correlate well with intelligence and aptitude tests.

Another line of research since 1992 has pioneered the use of careful measures of the timing of responses in working-memory tasks. We have shown that younger children take longer to search in short-term memory of a list for the word to be recalled next. This search rate, along with rehearsal rate, accounted for an astonishing 87% of the age-related variation in digit span (Cowan et al., 1998, Journal of Experimental Psychology: General). We recently used response-timing measures to show that different versions of a commonly-use working-memory procedure involve very different mental processes. If one must count displays of objects while remembering the sums from each display, attention must be shared between both tasks. However, if one must comprehend sentences while remembering the final word of each sentence, it is possible to switch attention away from the words and then reconstruct memory for them by reflecting upon the sentence material. One can tell because the latter type of process takes much longer (Cowan et al., 2003, Journal of Experimental Psychology: General).

Other interests. Much of my work has focused on the nature of acoustic and phonetic representations of sounds and how they automatically persist in short-term memory. This automatic processing is like the other side of the coin compared to attention-demanding processing. There is an acoustic afterimage of sound that lasts about 1/4 second, followed by a longer, vivid memory of sound that lasts for several seconds. I reached this conclusion in a critical review of research (Cowan, 1984, Psychological Bulletin).

Acoustic afterimages are key for perception of coherent speech. One question of fundamental interest was whether infants perceive sounds in the same way as adults. In adults, auditory afterimages can be studied by presenting two very brief sounds in succession. If the sounds' onsets are closer together than about 1/4 second, the second sound tends to interrupt the identification of the first by interfering with the use of the first sound's afterimage. During my graduate-school years, colleagues and I devised a comparable test in preverbal infants (Cowan et al., Child Development, 1982). We could tell if infants could detect sound variations (e.g., the repeating pair "ah-ah" changing to "eh-ah") by measuring increases in a learned sucking response that allowed access to the variations. If the variations occurred within pairs of vowels separated by 2/5 second, infants could make the discrimination; but not within pairs separated by only 1/4 second. The implication is that infants use an auditory afterimage longer than adults. Researchers continue to use concepts from this work, and sometimes still use our version of the discrimination procedure. Acoustic memory develops also in childhood. We show that young children lose acoustic information about unattended spoken words much more quickly than older children during a 5-second retention period (Cowan et al., 2000, Journal of Experimental Child Psychology).

I also published a study (1987, Journal of Experimental Psychology: Human Perception and Performance) clarifying how adults use auditory afterimages. Subjects were to judge the loudness of brief tones. When two tones were presented close together, the first tone was judged quieter than when it was presented in isolation. This supports the concept that ordinary perception requires information from acoustic afterimages.

Finally, we have questioned how acoustic memory is lost. I have addressed this issue in work with measures of both behavior and brain electrical activity (e.g., Cowan et al., 1992, Journal of Memory and Language; 1993, JEP:LMC; Winkler et al., 2001, Journal of Cognitive Neuroscience). The primary source of short-term forgetting proves not to be the loss of information because time has elapsed but, rather, difficulty in retrieving information that no longer is perceived to be relevant to the most recent stimulation. This is a fundamental change in understanding.

Practical applications. It is often said that there is nothing as practical as a good theory. Our basic research ideas have been applied to improve the understanding of mental disabilities, in published collaborations with clinical researchers. Research on children with specific language impairment has shown that the short-term memory problems of these children occur because they are poor at using information about the serial order of spoken words, and because they are poor at converting printed items into a phonetic form to assist in remembering them (e.g., Gillam et al., Journal of Speech, Language, & Hearing Research, 1998). Research on schizophrenics has shown that these patients form imprecise mental representations of sounds, though they do not have trouble remembering these sounds once the representations are formed (e.g., March et al., Journal of Abnormal Psychology, 1999). This finding contradicts a widespread view that the damage in schizophrenia is restricted to the frontal lobes, which are responsible for planning and remembering information. Research on amnesic patients (victims of stroke and head injury) has shown that their memory is much improved when there is no stimulation for up to 1 hour between the presentation and recall of story information (e.g., Della Sala et al., Memory, in press), opening up exciting possibilities for memory training in these patients.


[ Research ] [ Working-Memory Lab ] [ Publications ] [ Teaching ]
[ Student Opportunities ] [ Resources ] [ C.V.] [ Home ]

[ Department of Psychological Sciences ] [ College of Arts and Science ]
[ University of Missouri-Columbia ]



Revised: January 2016

Copyright © The Curators of the University of Missouri

Web Credits