The concept of selective attention is primarily characterized in terms of an ability to select, i.e., an ability to respond in a predetermined way to only one from a number of equally potent stimuli. The operation of selective attention can be demonstrated by a subject’s ability to select information with perfect accuracy. In short, selective attention selects and is very good at that.
Selective attention manifests itself rather by producing temporary sensory mechanisms or functions for selection of specific, relevant, stimuli; these selection mechanisms are called attentional traces. Essentially, these are voluntarily maintained neuronal stimulus representations corresponding to the physical features of the stimulus to be attended. These traces, located in the sensory-specific areas, explain the high speed and efficiency of a stimulus-set kind of selective attention and performance. The attentional-trace system is biased toward recognizing a certain stimulus, as if tonically asking whether an incoming sensory input is this stimulus or not. Hence, the sensory systems additionally perform task-related sensory analysis, as opposed to the task-unrelated, stimulus-determined, sensory analysis that the sensory systems are continuously performing irrespective of the direction of attention. However, the sensory neuronal populations participating in the two forms of processing are probably somewhat different.
The concept of selective attention is, however, immediately brought into relation with concepts regarding the capacity of the information processing system. The operation of selective attention is necessary because a subject has only a limited capacity to process, i.e., to identify or categorize, the information, and perfect selection is necessary, otherwise capacity is wasted. If there were unlimited capacity for processing information there would be no necessity for selective attention. All stimuli would then be processed.
An assumption, not explicit in the quotation, but nevertheless strongly present in the background, is that it is actually attention that is responsible for the processing, and that it is actually attention that is limited in capacity. So, lurking in the background is the ‘reading’ assumption, combining, within the concept ‘attention’, the concepts ‘selection’ and ‘processing’. In fact, this assumption was the dominant assumption of those days.
The attentional trace is thus suggested to be the physiological mechanism that subserves stimulus-set kind of selection. The processing-negativity data reviewed demonstrate, however, that the fast discrimination performed by this mechanism is much more accurate than what appears to be consistent with the previous views of stimulus set, which suggest the latter to operate only on rough physical differences. Moreover, whereas those views imply termination of processing of rejected stimuli, the present view holds that their processing is completed with regard to the basic physical attributes, selection involving only entry to temporary task-related processing stages maintained for task performance.
Selective attention determines the target for smooth eye movements. Involuntary mechanisms in the smooth oculornotor subsystem do not automatically average the velocity of all retinal motions. The role of involuntary mechanisms is limited to determining the velocity of the eye once a target is chosen by the subject.
Kahneman and Treisman (1984) have recently pointed out that the interest in the behavioral adaptive function of selective attention has progressively shifted, during the last 10 years, from perceptual processing to motor processing. They especially stress that this shift is closely related to a crucial paradigmatic change. Initially, in the framework of a limited-capacity model of attention, what they called the “filtering paradigm” appeared increasingly unsuccessful in localizing a privileged site for selection in the information-processing sequence. In the framework of a central or modular attentional processor’s flexibly allocating processing resources to different processing stages, what they called the “selective-set paradigm” emphasized the functional role played in goal-directed behavior by controlled and automatic processing, respectively.
One of the major changes that have occurred in attention research in the last few years is the increasing interest in selective set in motor control. This emphasis makes the division between sensory and motor processes confusing. For instance, the distinction between passive and active selective attention should be replaced by emphasis on ensuring the adequate execution of an intended action by an appropriate motor set, which results in an appropriate sensory set. Their basic finding is that visual-related neurons of the superficial layers of the superior colliculus respond to a visual stimulus in their receptive field more strongly when the stimulus is a target for a saccadic eye movement than when the monkey maintains its gaze on the fixation point.
The fact that this example is drawn from the field of behavioral neurosciences is not without significance. Because the selective aspects of information processing often are explained in terms of either controlled or automatized set, it is useful to link these concepts with those developed in the study of functional brain mechanisms underlying goal-directed behavior. Preparation is one of these concepts. However, its relations with selective attention, as well as the status of both attentional processes and preparatory processes in information-processing models underlying motor behavior, have yet to be specified.
In its most common usage, the notion of preparation is mainly associated with motor activity, and covers the set of processing operations that intervene before movement execution — that is, response selection or determination, and movement planning or programming. In this sense, motor preparation is viewed as a stage, or a set of stages, in current models of information processing. It is thus different from attention. Attention is never considered as a processing stage, but as a process, outside of the information-processing sequence, that changes the functional features of one or more stages, giving priority to a specific input or output. In this context, however, the fact that one specifies preparation as “motor” implies, of course, that it also can be specified in other ways — for instance, as sensory or perceptual (Posner, 1978 ). However, when preparation is specified in this manner there is an obvious shift in its meaning. Perceptual preparation is never used as an equivalent of one or several input-processing stages, but clearly covers other kinds of processes, mainly selective set.
Selective Attention Measures
When it comes to analyzing percepts, it would be convenient if we could drop them on the floor and watch them shatter into natural pieces. The stabilized image technique is probably the closest we have come to this so far. In the absence of such a direct technique, more indirect alternatives must suffice. When Garner and Pomerantz (1973) first approached the problem of grouping and configuration in vision, the dissecting tool they chose was the selective attention task. As this paradigm (and the logic behind it) is so well known from Garner’s papers on stimulus integrality and perceptual independence, Pomerantz described it only briefly. The subject is presented with stimuli that vary in two of their component parts or dimensions. The task is to classify those objects according to one part while ignoring the other part. The logic of this procedure is that if the two parts in question are dissected into separate perceptual units, then selective attention to just one part should be possible. But if the two parts are parsed into the same perceptual unit or group, then the two should not split, and so selective attention to just one part should be difficult or impossible. In this manner, the success or failure of selective attention becomes an operational measure of perceptual grouping. (Kahneman, 1973)
In conclusion, the mechanism of selective attention and general arousal are similar, the one resulting from a selective organization of thalamic gating mechanisms, and the other resulting from a nonspecific opening of all of the thalamic gates. As a result of such a comparative evaluation, significant stimuli are selected as requiring necessary perceptuomotor responses. This type of selection involves response set attention. Such response sets can in part be primed by activation of the cortical regions that evaluate stimulus—response contingencies, this activation being recorded as a localized negativity over the cortex specific to the response. The late positive wave of the evoked potential occurs in relation to the final response selection and in general reflects the amount of unequivocal information processed from stimulus to response. The negativity recorded mainly from the frontal regions of the brain during attention probably reflects the neural processes underlying the organization of attention and the allocation of effort to those aspects of sensory information processing that are significant to the underlying goals and needs of the subject. Such processes select the source of incoming information and set the context of its processing.