Attention and Memory
ATTENTION AND MEMORY
It seems to be a tenet of ordinary common sense that people remember what they attend to and forget what they do not. Not surprisingly, researchers have noted the very close relationship between attention and memory for a very long time, and some empirical evidence for the linkage was offered as far back as the late nineteenth century (Smith, 1895). However, it was only during the twentieth century, with the advent of cognitive psychology and its relatively rich array of methods for studying human information processing over fine time scales, that it became possible to begin to analyze this connection in more detail. To do so, researchers have used taxonomies of memorial and attentional processes that emerge from laboratory studies in each of these areas.
Forms of Attention
The term attention as used in everyday language is a diffuse and global term that alludes to both selectivity and capacity limitations. The potential for selectivity is evident in the fact that of the great multitude of stimuli impinging on the sense organs at any one instant, human beings are usually vividly aware of only a fairly small subset. Capacity limitations are evident whenever people try to attend to more than one stream of inputs, particularly if comprehension or response is required: for example, trying to listen to the radio at the same time as one reads a newspaper. Casual usage seems to imply that attention refers to a single process or substance that accounts for both selectivity and capacity limitations. Thus, people speak of focusing attention on one sensory input or one task, with the result that one does not have enough attention left over for other inputs or other tasks. Ordinary usage also seems to imply that limitations of attention are a key factor restricting one's ability not only to perceive many stimuli at once, but also to perform two tasks at the same time. These commonplace notions from folk psychology may or may not accurately describe humans' information-processing machinery.
Late-twentieth-century research argues that the phenomena of attention are not as unitary as common sense might suggest (Pashler, 1998). This is scarcely surprising given the complexities of the body's underlying neural machinery and of the tasks to which the human brain is adapted. Evidence for distinct attentional mechanisms is seen most clearly in relation to divided attention: the performance limitations that arise when a person attempts to process more than one stimulus at a time. On the one hand, there appear to be a set of processing limitations associated with perceptual analysis of inputs in different sense modalities. If a person is confronted with different sensory inputs at the same time, it is more difficult to perceive them when they arrive through the same sense modality rather than through different modalities. This was first demonstrated by A. Treisman and A. Davies (1973), who showed that people were better able to monitor animal names when some of the words to be monitored were presented visually and others auditorily (as compared to both in the same modality). These modality-restricted perceptual capacity limitations are wholly governed by the voluntary direction of selective attention: It is only the stimuli that are attended to that compete for limited resources, not all the stimuli that may be impinging on the senses.
In addition to the perceptual processing limitations tied to particular input modalities, research points to a separate set of limitations that become evident only when a person tries to perform multiple tasks at the same time. These central attentional limitations have their locus in the more cognitive stages of processing, especially the planning of actions and the retrieval of information in memory. When two tasks each require mental operations of these types, the processing in the two tasks is normally subject to queuing: selection of a response in one task must be completed before selection of the response in the other task can commence (Pashler and Johnston, 1998). Interestingly, these limitations seem quite indifferent to what modality the information arrives in. If a person must make a speeded response to a tone, for example, this will delay his or her ability to make a rapid response to a concurrently presented color patch. The central interference is also independent of response modalities; for example, if one response is vocal and one manual, the interference is still observed.
When one seeks to understand the relation of attention to memory, it is fruitful to inquire both about the role of modality-specific perceptual attention mechanisms and the role of central attentional mechanisms.
Forms of Memory
Memory, too, appears to be composed of a number of relatively separate functional systems. Several key distinctions have been proposed since the mid-1960s. The most important and well validated of these is the distinction between three broad memory systems that hold information over different time scales (and differ in certain other properties). This analysis, sometimes called the "modal model," postulates separate sensory memory, short-term memory (STM), and long-term memory (LTM) systems. The model arose out of the pioneering work of N. Waugh and D. A. Norman (1965) and M. Glanzer and A. R. Cunitz (1966). Sensory memory systems seem to produce something akin to a brief "literal" persistence of a stimulus for a very short time beyond its actual presentation. This probably reflects continued firing of neurons in sensory pathways after the offset of a stimulus. Short-term (or working) memory refers to a set of very limited-capacity storage mechanisms. It is often assumed to reflect an actively refreshed neural representation in sensory, perceptual, and motor-control areas, perhaps maintained by a reverberative process involving frontal brain structures. Long-term memory, on the other hand, refers to the relatively more permanent set of memory traces that is almost surely encoded by changes in synaptic weights. In addition to the three-part demarcation, many investigators propose a distinction within the realm of long-term memory, distinguishing between so-called explicit memory (underlying conscious recollections, as in recall or recognition tasks) and implicit memory (e.g., changes caused by exposure to a stimulus that modulate later processing without causing conscious recollection).
Attention and Sensory Memory
Sensory memory systems can potentially retain a large amount of detailed sensory information about an input for an extremely short period. For visual inputs, the sensory memory is called iconic memory. Iconic memory seems to hold onto inputs for only about 100 to 400 milliseconds, depending on physical properties of the stimulus and the visual input that follows it. Sensory memory for auditory sensations is usually called echoic memory, and most investigators agree that information is normally retained in echoic memory for one to two seconds. Sensory memory systems, while impressive in capacity, do not retain information long enough to be useful for most purposes. If the information is not transferred to short-term and/or long-term memory, it is lost.
Does information get into sensory memory even when a person attempts to ignore it at the time it is presented? For auditory sensory memory, the answer is evidently yes. This is seen, for example, in early observations by Broadbent and others that when people shadow spoken input to one ear, they can, if interrupted, abruptly switch over and (relying on echoic memory) recall the last bit of information presented to the other ear. Daily life offers many examples of the same phenomenon. Although there is little research that directly addresses the question of attentional involvement in iconic memory storage, the data that do exist suggest that even unattended information is briefly represented in this literal memory storage.
Attention and Short-Term Memory
Primary or short-term memory maintains information in an active state, but only so long as it is at least periodically rehearsed or refreshed. While some theorists originally proposed a single unitary STM, it has become clear that there are a number of separate and independent short-term memory systems. One system (sometimes termed the articulatory buffer and familiar to everyone who has ever remembered a phone number for a half minute or so) holds speech-like representations of verbal material. Another kind of STM system holds visual information in a schematic form. There are hints that other independent systems may also exist, holding, for example, nonvisual spatial representations, acoustic imagery, semantic representations, and tactile or kinesthetic patterns. The clearest evidence that different STM systems are indeed independent systems comes from experiments showing that people can hold onto more than one type of information at the same time. For example, people can retain spoken digits and visual materials with little mutual interference. Further clinching the case is the finding that some patients, after suffering brain damage, have lost one form of STM without any loss of the others (Basso, Spinnler, Vallar, and Zanobio, 1982).
Many researchers refer to "working memory" to indicate that information in STM may be actively manipulated or transformed during the time it is retained. Whatever purpose information is put to and whatever term may be used for it, the most notable characteristic of STM storage is limited capacity. For spoken material, a very limited number of words (or perhaps more critically, syllables) can be retained. For visual patterns, even a four-by-four checkerboard grid is enough to overload visual short-term memory. Access time for STM is quite good, however. For example, it appears that the image of a letter can be transferred into visual short-term memory in a matter of a few hundred milliseconds, and the rate for speech is probably not much different. What is the connection between attention and STM storage? Attending to a stimulus is clearly a necessary condition for storing that stimulus in short-term memory. Less clear is whether attending to a stimulus is a sufficient condition. The critical experiment would involve asking whether someone can retain one set of information (A) in STM and then attend to additional information arriving in the same modality (B) without having B overwrite A. Using visual patterns, W. A. Phillips and F. M. Christie (1977) concluded that this may be possible but that the issue requires further study. As for storage of spoken information, attending to irrelevant speech often compromises short-term memory quite considerably, suggesting that mere attention to a new input overwrites the existing contents of STM.
Does the storage and retention of information in STM require central attentional mechanisms discussed above? For articulatory STM, the involvement appears to be intermittent rather than continuous (perhaps arising in some initial consolidation in STM, and then later in scheduling periodic rehearsals). A person can retain a phone number, for example, and perform another brief demanding task involving unrelated materials without losing the phone number. However, if the demanding task is initiated immediately after storing some spoken material, memory for this material may suffer (Naveh-Benjamin and Jonides, 1984), and the same can happen when people undertake a continuously demanding task for some sustained period.
The picture that emerges, then, is of short-term memory systems closely tied to perceptual input systems (and corresponding perceptual attention machinery). Central attentional limitations may have some involvement, but it is relatively indirect or intermittent.
Attention and Long-Term Memory
To store information in long-term memory, one need not do anything active to maintain it. Nonetheless, recollecting a memory tends to strengthen its long-term memory representation, often dramatically so; conversely, memories are probably subject to erosion with the mere passage of time. What seems to be most difficult is getting information into LTM, and getting it out (retrieval). Devoting perceptual attention to a stimulus is by no means sufficient for LTM storage (whereas it may be sufficient for STM). A person can see one hundred words exposed one at a time in rapid succession (at a rate of, say, four per second), successfully detecting every word in the list that is an animal name, and at the end he or she will often remember scarcely anything except animal names. Obviously the person attended to all the words in order to determine their semantic category, but the process left no permanent residue.
What does produce memory storage, then? Many studies have concluded that the key factor is elaboration: active processing that uncovers connections between a to-be-stored item and other information in long-term store (Craik and Lockhart, 1972). By contrast, mere intention or desire to remember seems relatively unimportant.
Whereas central attentional mechanisms may play only a fairly minor role in the encoding of information into STM, for LTM they appear critical. Nearly every study that has looked at people's ability to form long-term memory traces while performing a concurrent task has found a substantial impairment. This occurs even when there is no discernible similarity between the stimuli to be remembered and those used in the concurrent task: for example, remembering odors while playing a computer game (Perkins and Cook, 1990).
In the late 1990s there was a lively controversy about the role of central attentional mechanisms in retrieval of information from long-term memory. On the one hand, when people are given unlimited time to retrieve materials, a concurrent task often fails to much dent the number of items they can produce (Craik, Govoni, Naveh-Benjamin, and Anderson, 1996). On the other hand, a demanding concurrent task markedly reduces the rate at which information can be retrieved from LTM (Hicks and Marsh, 2000). Furthermore, when an unrelated speeded-choice reaction-time task is performed concurrently with a paired-associate retrieval task, there is evidence that the memory retrieval is completely delayed by central attention to the speeded task (Carrier and Pashler, 1995).
Attention and Implicit Memory
The idea of a separate implicit memory system grew out of data showing amnesics may have a spared ability to form certain kinds of memories: those for which explicit recollection is not required (or at least a subset of such memories). A number of studies reported that formation of implicit memories also does not require limited-capacity attentional resources. For example, M. E. Smith and M. Oscar-Berman (1990) observed that a concurrent task at the time of word encoding did not reduce the priming effect found for repeated words as tested in a later lexical decision task. However, more recent studies tend to find that there are considerable costs when more demanding secondary tasks are performed at the time of encoding (Mulligan and Hornstein, 2000; Rajaram, Srinivas, and Travers, 2001). Researchers have more to learn about this conflict. It is possible that explicit memory measures are simply more sensitive to impaired storage produced by divided attention, but it is also possible that implicit memory storage reflects mechanisms that are independent of central attentional capacity.
Basso, A., Spinnler, H., Vallar, G., and Zanobio, M. E. (1982). Left hemisphere damage and selective impairment of auditory verbal short-term memory: A case study. Neuropsychologia 20, 263-274.
Broadbent, D. E. (1957). Immediate memory and simultaneous stimuli. Quarterly Journal of Experimental Psychology 9, 1-11.
Carrier, L. M., and Pashler, H. (1995). Attentional limits in memory retrieval. Journal of Experimental Psychology: Learning, Memory, and Cognition 21, 1,339-1,348.
Craik, F. I. M., Govoni, R., Naveh-Benjamin, M., and Anderson, N. D. (1996). The effects of divided attention on encoding and retrieval processes in human memory. Journal of Experimental Psychology: General 125, 159-180.
Craik, F. I. M., and Lockhart, R. S. (1972). Levels of processing: A framework for memory research. Journal of Verbal Learning and Verbal Behavior 11, 671-684.
Glanzer, M., and Cunitz, A. R. (1966). Two storage mechanisms in free recall. Journal of Verbal Learning and Verbal Behavior 5, 351-360.
Hicks, J. L., and Marsh, R. L. (2000). Toward specifying the attentional demands of recognition memory. Journal of Experimental Psychology: Learning, Memory, and Cognition 26, 1,483-1,498.
Mulligan, S., and Hornstein, S. (2000). Attention and perceptual priming in the perceptual identification task. Journal of Experimental Psychology: Learning, Memory, and Cognition 26, 626-637.
Naveh-Benjamin, M., and Jonides, J. (1984). Maintenance rehearsal: A two-component analysis. Journal of Experimental Psychology: Learning, Memory, and Cognition 10, 369-385.
Pashler, H. E. (1998). The psychology of attention. Cambridge, MA: MIT Press.
Pashler, H., and Johnston, J. C. (1998). Attentional limitations in dual-task performance. In H. Pashler, ed., Attention. Hove, East Sussex, UK: Psychology Press.
Perkins, J., and Cook, N. M. (1990). Recognition and recall of odours: The effects of suppressing visual and verbal encoding processes. British Journal of Psychology 81, 221-226.
Phillips, W. A., and Christie, F. M. (1977). Interference with visualization. Quarterly Journal of Experimental Psychology 29, 637-650.
Rajaram, S., Srinivas, K., and Travers, S. (2001). The effects of attention on perceptual implicit memory. Memory & Cognition 29, 920-930.
Smith, M. E., and Oscar-Berman, M. (1990). Repetition priming of words and pseudowords in divided attention and in amnesia. Journal of Experimental Psychology: Learning, Memory, and Cognition 16, 1,033-1,042.
Smith, W. G. (1895). The relation of attention to memory. Mind 4, 47-43.
Treisman, A., and Davies, A. (1973). Dividing attention to ear and eye. In S. Kornblum, ed., Attention and performance IV. New York: Academic Press.
Waugh, N., and Norman, D. A. (1965). Primary memory. Psychological Review 72, 89-104.