Conflict, Target Detection and Cognitive Control

Michael I. Posner and Gregory J. DiGirolamo

University of Oregon


Abstract

According to cognitive models, executive control is required when tasks involve planning, error detection, novelty, difficult processing, or conflict. These are situations in which it becomes impossible to process more than one target event at a time and neuroimaging studies have found activation in frontal midline areas. We examine the anatomy and time course of executive control in therapid generation of novel word meanings and in Stroop conflict tasks. We review evidence that relates this higher level attention network to psychopathology and to cognitive development. We also consider states that might serve to dissociate components of the executive network.


This is a draft to appear in:

R. Parasuraman, (Ed.) The Attentive Brain Cambridge: MIT Press

I. Why Executive Control?

All normal people have a strong subjective feeling of intentional or voluntary control of their behavior. Asking people about goals or intention is probably the single most predictive indicator of their behavior during problem solving (Newell & Simon, 1972). The importance of intention and goals is illustrated by observations of patients with frontal lesions (Duncan, 1994) or mental disorders (Frith, 1992) that cause disruption in either their central control over behavior or the subjective feelings of such control. Despite these indices of central control, it has not been easy to specify exactly the functions or mechanisms of central control.

In this chapter, we first review efforts to develop a cognitive model of executive control and consider how experimental methods can be used to explore conditions in which executive control will operate. In our second section, we examine neuroimaging studies that employ these cognitive methods to explore the anatomy and circuitry of executive control. Our third section considers evidence from lesion, schizophrenia, and developmental studies that provide further tests of which areas are involved in executive control. Finally we consider future opportunities for dissociating components of executive function. It has always been hard to define executive control precisely because the term has been used in many ways. We try to avoid these difficulties by use of a specific model of higher level control to specify when executive control is a necessary aspect of information processing. Moreover, we believe some progress is being made in identifying components that might provide further analysis of the mechanisms that together constitute executive control.

The Norman-Shallice Model

Norman and Shallice (1980; 1986) have created one representative model of executive control that assumes multiple, isolable subsystems of cognitive processing. These multiple subsystems interact to coordinate goals and actions (see, Allport, 1980a, 1980b) and are controlled by two qualitatively different mechanisms. The first level of control operates via "contention-scheduling " which uses "schema's" or condition to action statements to coordinate well-learned behaviors and thoughts (c.f., Newell & Simon, 1972). Once a schema is selected, it remains active until it reaches its goal or it is inhibited by a competitive schema or higher-level control.


Figure 1 The Norman-Shallice Model (adapted from Shallice, Burgess, Schon, et al., 1989)

The contention-scheduling mechanism corresponds to routine selection. When the situation is novel or highly competitive (i.e. requires executive control), a "Supervisory System" would intervene and provide additional inhibition or activation to the appropriate schema for the situation (see Figure 1). The Supervisory System has access to the overall representation of the environment and the goals of the person, unlike the contention-scheduling mechanism which involves only competition among subsystems.

Norman and Shallice (1980; 1986) have argued that the Supervisory System would be necessary for 5 types of behaviors or situations in which the routine or automatic processes (Shallice, 1994) of the contention scheduling mechanisms would be inadequate and the executive control of the Supervisory System would be required: 1) Situations involving planning or decision-making; 2) Situations involving error-correction; 3) Situations where the response is novel and not well-learned; 4) Situations judged to be difficult or dangerous; and 5) Situations that require overcoming habitual responses.

We also suggest that executive control operates only in certain situations, or during portions of tasks, when these executive functions are necessary. Executive control operations are necessary when routine functions (subsystems) are insufficient for the task at hand, or when subsystems must be overridden due to environmental or goal changes. Executive control is not a continuous and universal process present in every cognitive activity; though these cognitive operations may play an important part in the unification of thought and action (see, Allport, 1980a, 1980b). We further suggest that executive control operations are dissociative from other cognitive operations in a task (see Rogers & Monsell, 1995). In this chapter, we will discuss some evidence of brain systems involved in situations calling for executive control.

Means of Control

In the Norman-Shallice model, contention-scheduling works via local inhibition of competing schemas. In a recent review of the neuroscience literature, Desimone & Duncan (1995) have also argued that visual sub-processing systems perform selection via a local competition in which the receptive fields are viewed as critical resources for which stimuli are competing. Desimone & Duncan believe that the competition for these sub-processing systems is resolved via inhibition in the local neural circuit.

In addition, Desimone & Duncan argue that the competition can be biased by a top-down mechanism that selects objects that are important to the current behavior or goal. Several studies have found preferential increases of neuronal activity (Motter, 1993; Spitzer, Desimone & Moran, 1988), or of blood flow (Corbetta, Miezin, Dobmeyer, Shulman & Petersen, 1991) for selected features or locations. Scalp electrical recording has demonstrated both increases of electrical activity from selected events and decreases from competing event when compared to a neutral condition (Luck, et al., 1990). Like the Supervisory System mechanism of the Norman-Shallice model, competition for control of behavior appears to be resolved at local sites by the relative amplification of the selected competitor.

Interference between targets

In order to understand the mechanisms of supervisory control it is necessary to have reliable experimental techniques for causing executive control to be employed. A well established principle of cognitive psychology is that there will be interference whenever two tasks require access to the same underlying systems. Interference between tasks that use quite separate input and output pathway has been taken as a method for the measurement of central attentional control.

Early work in attention presented two separate streams of information, one to each ear, and required subjects to rivet their attention to one stream by repeating it back (shadowing) (Broadbent 1958, 1973). Most information from the non shadowed ear is lost. In a basic experiment of this era, Treisman and Geffen (1967) asked subjects to tap a key whenever they heard the word 'tap'. When it occurred in the attended ear they tapped the key close to 100% but when it occurred in the unattended ear they almost never did. On the other hand, it has been shown that significant events on the unattended ear or presented visually during shadowing, can still produce priming or galvanic skin response (Corteen & Wood, 1972; Corteen & Dunn, 1974; Dawson & Schell, 1982; Posner, Sandson, Dhawan, & Shulman, 1989).

When the requirement to shadow one of the two messages was relaxed (Duncan, 1980; Ostry, Moray, & Marks, 1976; Shiffrin, McKay & Shaffer, 1977) and instead the experimenter required subjects to monitor 1 or N channels, accuracy of detection of the target remained very high regardless of the number of channels. In opposition to the shadowing results, subjects appeared to have nearly unlimited capacity to monitor incoming messages. A theoretical account by Duncan (1980) helped to reconcile the shadowing and monitoring studies. He showed attention could be summoned to one of several input channels with very great effectiveness, but when a target occurred on one channel, processing of targets on any other channel dropped dramatically. Even if subjects only think a target has been present and they make a false report, performance on simultaneous signals greatly deteriorated (Ostry, Moray, & Marks, 1976). Thus, major interference is found among items selected for focal attention (targets).

This important insight continues to be rediscovered in various domains. When events are presented in rapid succession (Raymond, Shapiro & Arnell, 1995; Shapiro, Raymond & Arnell, 1994), attending to one target appears to reduce the ability to detect a second target for .5 second (attentional blink). Although in the attentional blink experiments there is a stream of information presented at a rapid rate, studies of the psychological refractory period indicate that even in self paced tasks if attention is given to the processing of one signal, new signals that appear prior to the response are delayed (Pashler, 1993; Pashler & Johnston, 1989).

Some of these ideas have been applied to switching between successive tasks somewhat more in line with what might be the function of an executive mechanism (Allport, Styles & Hsieh, 1994; Rogers & Monsell, 1995, Specter & Biederman, 1976). In this paradigm subjects perform one task (e.g. naming ink colors) and then switch to a different task (naming the value of a digit). The switch might be signaled by a change of stimulus or the same stimuli might appear and the task may be alternated. Irrespective of the nature of the task, subjects appear to be slower on the first trial following a switch even when the task is paced so that no new event is presented until after the previous event has been responded to (Rogers & Monsell, 1995). Thus there is a clear cost in switching between tasks. However, it is less clear that the cost is due to an operation involved in switching set; rather, there appears to be some residue of having done one task that makes processing the new task more difficult.

Overall these results show that the difficulty in processing simultaneous or successive targets extends across a wide range of tasks. Apparently something about having to commit attention to an event interferes with processing other events at this level irrespective of the similarity between the two tasks. This observation supports the idea of some kind of central process that is common to a wide range of targets. The problem remains of how to define targets? It is simple if the experimenter identifies some event for special processing, as when a key press must be made to detect, discriminate, identify or classify a stimulus. However, events may serve as targets because they automatically force processing. This may happen with a transient visual onset stimulus that forces orienting (Yantis & Jonides, 1990), or when a distracter with a competing classification is placed close to a target (Eriksen & Eriksen, 1974). Often the tendency to read the name of a visual word will interfere with other more arbitrary responses to that word such as the ink color (Stroop effect) or providing associations such as its use or its superordinate category. The next section uses these target interference effects to explore the anatomy of executive attention.

II. Imaging Executive Function

Anatomy

Tasks requiring supervisory control (Duncan, Burgess, and Emslie, 1995; Shallice, 1988 Shallice and Burgess, 1991a; 1991b) are severely affected by lesions of the frontal lobes. However, these tasks were quite complex with many components and the frontal lobes are a large part of the brain, so here we review studies using neuroimaging methods to provide better localization of supervisory functions.

The methods of neuroimaging allow examination of brain anatomy during task performance (Corbetta, this volume). Many studies involving the detection of targets or the resolution of conflict among targets have found activation in a midline frontal area called the anterior cingulate. We concentrate on two of these tasks: the generate uses task (see Petersen, Fox, Posner, et al., 1988; Petersen, Fox, Posner, et al., 1989; Posner, Petersen, Fox, et al., 1988) and the Stroop task (see MacLeod, 1991 for a review; Stroop, 1935).

The generate uses task involves a experimental condition of giving the use of a familiar visual or auditory word and a control task in which the subject merely reads or says aloud the word name. Blocks of 40 trials were used during which blood flow was averaged. The subtraction of generate - repeat revealed three cortical areas of activation. There were activations in the anterior cingulate and two left lateral areas, one in a frontal area anterior to Broca's area and one in Wernicke's area. The lateral areas were near classical language areas. The midline area seemed more likely related to attention to the task.

What happens when subjects practice generating the same use over and over again? The Norman Shallice model would predict that following practice, a "schema" would be formed that would trigger when the stimulus was presented and the Supervisory System would not be necessary. Raichle et al., (1994) had subjects perform the Generate uses task while being PET scanned and found the expected activation of the anterior cingulate when compared to repeating the word aloud. In addition, Raichle had subjects practice the same list repeatedly, generating the same appropriate use for each word until the list was highly learned. Following the extended practice, subjects again were scanned. This time, the anterior cingulate and the left lateral activation were gone, instead there was increased activation in the anterior insula which now was similar to the activation found when reading a word aloud. Following the practice, Raichle et al. had subjects generate a use for a new, unpracticed list. Again, the anterior cingulate and left lateral areas were active. Thus, the anterior cingulate is active when the Supervisor System is held to be necessary for appropriate behavior in the Norman-Shallice model, and the anterior cingulate is inactive when the Supervisory System should be inactive and the contention-scheduling mechanisms active. However, the cingulate was not alone, the lateral activations behaved in a similar manner. This could be because attention increased the lateral activations and when attention was reduced they fell below threshold.

The Stroop tasks involve the naming of the color of ink of a word which can be congruent (matching the color of the ink it is printed in; e.g., the word "RED" in red ink); neutral (non-color related; e.g., the word "LOT" in red ink); or incongruent (a mismatch of the word with the color of ink the word is printed in; e.g., the word "RED" in blue ink). One computational breakdown of the Stroop task can be outlined as follows: (a) remember instructions of vocalizing the ink color, (b) attend to the visually presented stimulus, (c) determine the ink color of the word, (d) inhibit the naming of the word, and (e) make the appropriate response (being careful not to name the color word presented). Anterior cingulate activation is confirmed by four separate Stroop (or Stroop-like) studies (see, Bench, Frith, Grasby, et al., 1993; George, Ketter, Parekh, et al., 1994; Pardo, et al., 1990; Taylor, Kornblum, Minoshima, et al., 1994), although the specifics of the studies may account for the varied locations of the activations (see Figure 2).

The first Stroop PET study was conducted by Pardo and his colleagues (1990) as a means of testing the neural areas specifically involved in selective attention and cognitive control. Each subject always began with the congruent condition and then went on to the incongruent condition at the next scan. The resulting imaging data was obtained by subtracting the congruent condition from the incongruent condition. The anterior cingulate was active in three areas on the right medial portion and one area in the left inferior sulcus. The averaging necessary for the PET methodology requires that experiments be blocked in one task of the same condition for 40 secs followed by a brief rest, and the next condition of the task for another 40 secs. The resulting paradigm is problematic for the Stroop task as one cannot know if in the congruent condition the subject is deciphering the ink color or merely reading the word. Pardo and his colleagues always had the subjects perform the congruent condition first. Since the subjects first exposure to the task is the congruent condition, there is no necessity nor "attentional set" (see, Perret, 1973) to name the ink color.

Given the computational similarity of the congruent and incongruent conditions (see below), it is surprising that activation of the anterior cingulate was detected using the subtraction that Pardo et al. employed. Bench et al. (1993) attempted to replicate the Pardo et al. (1990) findings using the PET methodology. Unlike Pardo et al., Bench and his colleagues used a faster presentation rate and three conditions: (a) naming the ink color of colored crosses; (b) naming the ink color of neutral words ("front," "back," "top," and "down"); and (c) naming the ink color of incongruent color word names. In the subtraction of the neutral words from the incongruent condition, there was no activation of anterior cingulate. Bench and his colleagues were so surprised by this finding that they conducted another study using Pardo et al.'s (1990) paradigm exactly; the incongruent-congruent subtraction produced no significant areas of activation anywhere in the brain. If one compares the computational processes in the congruent and incongruent condition, both require inhibition of the competing color word in order to respond to the ink color, thus the same processes are occurring in different degrees in the two conditions. Bench, et al (1993) report an increase in the right anterior cingulate during both the congruent and incongruent conditions in comparison to the crosses. This finding suggests that the effort to name the ink color and avoid the word can influence both the congruent and incongruent conditions in the same way. Bench and his colleagues also concluded that the frequency of stimulus presentation greatly effect the activation of anterior cingulate (more stimuli and a faster pace increase the blood flow to the cingulate). This result suggests that as attentional demands are increased in making a faster judgment, the anterior cingulate also increases its role in the task.

In yet another PET study of the Stroop effect, George et al. (1994) used three conditions: (a) number symbols in colored ink (#####); (b) incongruent color words in colored ink; and (c) emotional words in colored ink ("sad," "grief," "misery," and "bleak"). Unlike the previous studies, George and his colleagues had subjects name the ink colors of words in lists with the stimuli presentation being self-paced. They found increased activation in left midcingulate, left anterior cingulate, and a right midcingulate activation. George et al. (1994) interpreted these results by suggesting that different portions of the cingulate were involved in different computational tasks. The studies which are self-paced and emphasize the response selection seem to show a left midcingulate activation; in contrast, in studies which are externally-paced, the right anterior cingulate seems to be active.

George et al.'s commentary on self vs. externally paced stimulus presentation receives support from a Stroop-like study by Taylor et al. (1994). In this study subjects are taught to respond to single letters with another single letter; that is, in the congruent condition, subjects see the letter "J" and respond by saying "J"; in the incongruent condition, subjects see the letter "J" and respond by saying a different letter that they have been taught corresponds to "J" ("F", for example). Taylor et al. found activation of the left cingulate sulcus in a subtraction of congruent from incongruent. It is important to note that in the Taylor et al. study, the incongruent task is a different computation than the congruent task. The congruent task is computationally simple: the stimulus appears and one names it; there is no degree of incompatibility with this task. In the incongruent condition, one must keep from naming the letter presented and instead name the different letter that the subject was taught corresponds to this stimulus.


Figure 2 Activation of anterior cingulate in 4 Stroop or Stroop-like studies

The Norman & Shallice (1980; 1986) model suggested five situations in which the Supervisory System would be necessary for appropriate behavior or successful execution of a task or goal. The first of these points was circumstances involving internally planned, or voluntary actions. Using positron emission tomography (PET; see chapter X), Passingham and his colleagues (Colebatch, Cunningham, Deiber, et al., 1990; Deiber, Passingham, Colebach, et al., 1991) have shown statistically significant increased activation in the anterior cingulate in voluntary, planned arm and hand movements when compared to resting, learned sequence movements, and fixed sequence movements. Two other situations in which the Norman & Shallice model has held the Supervisory System to be necessary are situations that require overcoming habitual responses and situations where responses are not well-learned or contain novel sequences of actions. These factors fit the generate use and Stroop tasks. Another situation in which the Norman-Shallice model necessitates a Supervisory System involvement is in error-correction or detection ("trouble-shooting"), studies involving error detection will be considered in the next section.

Circuitry

In order for a brain area to perform a supervisory function it must influence widely distributed parts of the brain where computations related to the task are performed (Posner & Raichle, 1994). Anatomical studies suggest that the anterior cingulate, like many brain regions, has close contact with many other cortical areas (Goldman-Rakic, 1988). The cingulate connections to lateral frontal areas involved in word recognition, and posterior parietal areas involved in orienting are particularly strong.

Recently efforts have been made to trace the dynamics of these interaction by use of event related potentials (see Näätänen, 1992; Rugg & Coles, 1995 for reviews of the ERP methods). While It is a difficult task to determine the generator from a scalp distribution of electrical activity, if the generator is known from PET or fMRI studies it is much easier to evaluate whether the scalp distribution could come from that generator (see Heinze, et al., 1994 for an example of this methodology). The algorithms to relate a generator to the scalp distribution (Scherg & Berg, 1993) work best when fewer generators are involved. This method assumes that the scalp ERP arises from the summation of some number of sources ("dipoles") within the brain of different locations, orientations, strengths, and time-courses. Using a "best-fit" method, these sources are modeled until the solution of the model most closely matches the empirically obtained scalp electrical data. Thus in complex tasks, it is important to use a subtraction which isolates only a small number of brain areas.

One effort to do so is in the localization of a scalp negativity that follows making an error (Dehaene, Posner & Tucker, 1994; Ghering, et al., 1993). When subjects were aware of making an error in speeded tasks they showed a very strong negativity following the key press in a localized area over the mid frontal scalp. Further analysis using the Brain Electrical Source Analysis (BESA) algorithm (Scherg & Berg, 1993) showed this error negativity most likely came from the anterior cingulate. Errors can be either slips, incorrect execution of a motor program, or mistakes, selection of an inappropriate intention. The Norman-Shallice model would predict Supervisory System involvement in the recognition of an execution of an incorrect motor program, but not if the contention-scheduling had selected what was thought to be an appropriate response. Dehaene, Posner & Tucker, (1994) have shown that error negativity follows a slip, where the person knows the error, but not if they are unaware of the error (Tucker, Liotti, Potts, et al., 1994).

Studies of the generate uses minus repeat words task (described above) using ERPs have provided a means of integrating the anatomy found by PET with a specific time course (Snyder, Abdullaev, Posner & Raichle, 1995). Figure 3 shows scalp potential maps of electrical differences between these two tasks. This figure shows increased positivity of the generate over the repeat task that occurs very early over frontal sites and much later over left posterior sites.


Figure 3 Maps of the relative voltage differences between the generate uses and reading aloud at several time slices. The colors indicate the size of the voltage differences with the bright colors indicating greater positivity in the uses task and the dark colors greater negativity in the uses task. The early frontal positivity is related to activity in the cingulate and left lateral lexical semantic areas. The later left posterior positivity is related to activity in Wernicke's area.

Using BESA, it was possible to show that the early frontal activity could be fit best by a midline generator (presumably anterior cingulate) starting about 170 msecs after the visual word. This activation was presumably related to some kind of focal attention. The cingulate activation stayed present and was joined after 50 msec by a left frontal activation. At about 650msec a single generator was found presumably related to Wernicke's area activation.

Abdullaev & Posner (1995) took a further step in replicating the PET results. They had subjects generate uses for the same list several times. The left frontal and cingulate activations (as localized by the BESA methodology, described above) tend to go away following practice, but the activations were restored when a new list was presented or when subjects were required to generate a new use for the practiced words.

Functions

These experiments suggest that the anterior cingulate is active during tasks that require some thought, and is reduced or disappears as tasks become routine as in the case of reading words aloud or following practice generating the same use.

What is the cingulate activation actually doing? According to our analysis the cingulate is involved in producing the local amplification in neural activity the accompanies top down selection of items. It is easiest to understand this function in the domain of processing words. It is well known from cognitive studies that a target word is processed more efficiently following a related prime word (Posner, 1978). A portion of this improvement occurs automatically because the prime word activates a pathway shared with the target. However, another portion of the activation is top down because the attention to the prime leads the subject to expect a particular type of target. If the prime is masked or of low validity, the improvement in the target will be mostly automatic; however, if the prime is of high validity or if subjects are instructed to use the prime to think of another category, top down effects dominate. If the target is ambiguous (e.g. palm), the prime (e.g. tree) can lead to a single conscious interpretation that fits with both prime and target (see Simpson, 1984 for a review). We believe that the cingulate is responsible for these top down effects by providing a boost in activation to items associated with the expectation.

Anatomically, we see the cingulate in contact with areas of the left lateral and posterior cortex that seem to be involved in understanding the meaning of a given target word. Indeed the time course of activation of the cingulate (170 msec) and the left lateral frontal (220 msec) cortex found during the generate task support our speculation that attention interacts with the semantic activation pattern (Snyder et al., 1995).

Primes presented to the right visual field produce rapid activation of close semantic associates while primes presented to the left visual field act more slowly and activate remote associates as well. Blocking attention to the visual word by shadowing (Nakagawa, 1991) reduces semantic priming and produces a pattern priming pattern that resembles that found when the prime is presented to the left visual field irrespective of the visual field to which the prime is presented. This finding suggests a specific role of attention in producing the left hemisphere priming pattern. Similarly, primes that are masked and those not subject to specific attentional effect based on their meaning activate a wider range of associates than those which are unmasked. These findings show evidence of the specific role of supervisory control over semantic activation patterns.

In addition, we believe that cingulate activation plays a role in the voluntarily reactivation of brain areas that can also be driven automatically from input. Feature analysis of visual targets appears to involve right lateralized posterior parts of the brain. If subjects are instructed to examine a feature voluntarily, similar electrode sites are activated much later (Posner & Raichle, 1994). We believe that the cingulate is important to these activation patterns. We also believe that the cingulate activation is involved when elements of a thought are reordered in time. By increasing activation of the brain area that performs a specific computation one can change the time course of the organization of the component operations (Posner & Raichle, 1994).

III. Lesions, Schizophrenia, and Development

Lesions of the frontal lobe often produced disorganized or incoherent behavior (Duncan et al., 1995; Shallice, 1991a; 1991b). In neuropsychology, the dysexecutive syndrome following closed head injury, stroke, or degenerative disorders of frontal structures involves the loss of the ability to plan coherently, to solve problems, or to organize the routines of daily life. Patients suffering from this syndrome have difficulty with problem solving tasks such as the Tower of Hanoi in which planning ahead is an important component.

There has been some controversy about the importance of cingulate involvement in the loss of executive function. Large lesions of the frontal midline produced by strokes can have devastating effects on human behavior. Damasio (1994) who has studied may of these patients suggests:

"Before leaving the subject of human brain lesions, I would like to propose there is a particular region in the human brain where the systems concerned with emotion, attention and working memory interact so intimately that they constitute the source for the energy for both external action (movement) and internal action (thought, animation, reasoning). This fountain head region is the anterior cingulate cortex another piece of the limbic puzzle" (p. 71).

This observation comes from patients who show akinetic mutism, following strokes in the general area of the cingulate. These patients can orient to events but initiate little in the way of spontaneous behavior. One woman studied by Damasio (1994) recovered and when Damasio asked her what was going on during the period of time when she suffered from the brain injury and why she didn't initiate any behavior or communication, she said, "I really had nothing to say" (p. 73). The fact that there can be recovery after a brief period of akinetic mutism suggests that there is considerable distribution of executive function both within the cingulate and in other structures related to it. There is a history of work with cats and monkeys involving lesions of the cingulate (Kennard, 1954; 1955; Ward Jr., 1948) which produced results similar to Damasio's studies. Both cats and monkeys with extensive anterior cingulate lesions show the same lack of initiation of voluntary behavior or movement.

However, as more discrete cingulate lesions have been used to treat patients with pain or anxiety there has been little evidence of the gross loss of conscious control reported in the studies cited above1 (Corkin, Twitchell, & Sullivan, 1979; Ballantine, Levy, Dagi, and Giriunas, 1977). Perhaps this relates to the various areas of the cingulate that might be involved in higher order attention as illustrated by Figure 2. In order to see any deficits in surgical patients it has been necessary to study the specific tasks that have shown consistent activation of the anterior cingulate during imaging (Janer & Pardo, 1991).

Janer and Pardo (1991) tested a 34 year old, right-handed, college educated female one week before, and two weeks after bilateral anterior cingulotomy, 10 control subjects were also tested on the following three cognitive tasks: (a) Semantic monitoring task (decide if a word is a dangerous animal) (b) Generate uses task; and (c) Stroop task.

As one would expect, the patient showed an overall increase in reaction time following her surgery. She showed no significant increase in reaction times post operation when the semantic monitoring task involved no animals but when non-dangerous animals (e.g., cat) were present, she showed impairment following the cingulotomy and also made more errors for non-dangerous animals than normal controls.

The data from the verb generation task also revealed a significant difference between the controls and the patient following the cingulotomy. The controls in their second testing showed an improvement in naming a use of the noun; in contrast, the patient showed an increase RT for naming a use for the verb following the cingulotomy.

Another difference between the patient and the controls was the latter were able to detect their errors, sometimes even before they reached the output stage. The patient never made any pre-output hesitations symptomatic of detecting errors. Controls made 32 hesitation errors during their second testing, the patient never made a hesitation error following her cingulotomy. The patient never detected an impending mistake in her performance even though she made more errors than the controls. Again, the cingulotomy patient showed differences from the controls on a task in which ERP data has implicated the anterior cingulate (Dehaene, et al. 1994).

Janer and Pardo also tested their patient and controls on a Stroop task using the same methodology as the PET study done by Pardo and his colleagues (1990). Both congruent and incongruent conditions of the classical Stroop task (Stroop, 1935) were used. Both the control subjects and the patient performed faster in the congruent condition than the incongruent condition during both times of testing. The control subjects performed faster on the second testing of both the congruent and the incongruent condition. The patient performed equally at both times of testing in the incongruent condition. However, the patient showed a significant deficit in performance, following the cingulotomy, in the congruent condition. This results proves difficult to reconcile with cingulate involvement in conflict situations. There is, however, an interpretation of the Stroop task which might explain a deficit in the congruent condition following cingulotomy. Both the incongruent and the congruent condition require the same selective attentional mechanisms, but to different degrees. In the congruent task, one must selectively attended to the color of the ink. Since word naming can be an automatic process (Posner & Snyder, 1975), then the word naming in both conditions should occur and slow down processing. Hence, the patient's difficulty in naming the ink color in the congruent condition might support a deficit in selective attention suggested by the other tasks. Nevertheless, we are faced with the problem that the patient did not show a deficit in the incongruent condition following her cingulotomy. The results with this single patient clearly require further investigation of cingulate operations.

Schizophrenia

Some recent analyses of schizophrenia are based on ideas of dysregulation of the anterior cingulate (Early, Posner, Reiman & Raichle, 1989 a, b). This view began with observations that patients suffering from the early positive symptoms of schizophrenia had difficulty in attention toward the right visual field (Posner, et al., 1988). This finding has now been confirmed by others (Maruff, et al., 1995).

The same schizophrenic patients had difficulties in dealing with a single word made to conflict with a spatial location in a Stroop- like task. This result, and other language related difficulties in schizophrenics, suggested to us that the deficit was arising in the anterior cingulate, since the anterior cingulate served to exercise control over both language and shifts of spatial attention. We were also able to relate our specific findings with more general deficits in schizophrenic thought such as the tendency to attribute aspect of their thoughts to others and the difficulty in producing coherent sentences.

These early studies did not provide as much evidence as one would like, but more recent evidence has continued to suggest that the cingulate (see also, Dolan, Fletcher, Frith, et al., 1995) in conjunction with basal ganglia and lateral frontal cortex might be central to the positive symptoms of schizophrenia . Benes (1993) has reported analysis of the cingulate following the death of patients who suffered from schizophrenia. Figure 4 points out the central role for the cingulate in her view and also presents a cellular model in which the GABA input to the pyramidal cells were affected in the schizophrenic patients. Benes suggests that decreased inhibitory inputs and increased excitatory inputs reduces regulatory firing of the pyramidal neurons in the anterior cingulate.


Figure 4 The upper diagram displays cingulate connections to areas implicated in the psychopathology of schizophrenia including areas active in attention, working memory, and emotion. The lower figure shows a schematic of anterior cingulate pyramidal neurons in layers II and III in normal controls (left) and patients with schizophrenia (right). The schizophrenic circuit shows less inhibitory connections (GABAergic interneurons) to the neurons and more excitatory connections (Glutamatergic axons) to their dendrites (adapted from Benes, 1995)

Our previous work (Early, et al., 1989a, 1989b) shows how many of the disorders found in schizophrenic patients could arise from dysregulation of the control processes of executive attention (see also Frith, 1992). This dysregulation would be expected as the result of the cellular abnormalities that Benes has outlined.

Development

One place where supervisory control has been frequently invoked is in the study of working memory (Baddeley, 1986). When subjects explicitly recall events they draw upon supervisory mechanisms to search memory. In recent years a distinction between such explicit recall mechanisms and implicit memory has been important and one domain in which it has been examined is in the learning of spatial sequences. For example, Curran & Keele, 1993, have studied the learning of a fixed sequence of spatial positions by normal subjects. They distinguish between unambiguous associations, in which each stimulus always implies a single next location and context dependent associations, in which the nature of the association depends upon context. These adult studies (Curran & Keele, 1993) indicate that unambiguous associations can be learned, presumably implicitly, even when attention is diverted by a secondary task, but context dependent associations cannot be learned without focal attention. Recent PET data of these tasks suggest that implicit learning of the sequences involves mostly subcortical areas, but explicit learning involves higher cortical areas (Grafton, Ivry & Hazeltine, in press).

Clohessy (1994) reported that 4 and 10 month old infants could learn simple sequences of unambiguous association which they indicated by correct eye movements in anticipation of the next event in the sequence. The studies of four and ten-month-old infants also employed a complex sequence in which one of the associations was context dependent or ambiguous. In this sequence, infants were shown a sequence where the target moved from monitor 1 to 2 and then after returning to 1, moved to 3 (i.e. 1->2->1->3). Thus the association of the location that follows monitor 1 was dependent upon where in the sequence they were. The ability to learn ambiguous associations could then serve as a marker task for the development of the more complex forms of attention related to executive control by anterior structures. Clohessy found that infants of four and ten months learned to correctly anticipate the unambiguous returns to position 1, but showed no evidence of learning the context dependent association.

By eighteen months, infants are showing many signs of higher level attentional control. This includes the "language explosion", emergence of multiple word utterances, the ability to sort and classify objects and evidence of self recognition (Meltzoff, 1990). There was clear evidence that the eighteen-month old infants learned the context dependent association. Some infants showed by their correct anticipation that they had learned this skill very well and others showed little evidence of learning. This finding suggests that infants of this age are only beginning to acquire the ability to learn this skill, and that it makes worthwhile observing how this form of learning develops in somewhat older infants.

We were also able to show some links between the learning by infants at this age and aspects of their language performance in daily life. We found a significant correlation between their laboratory performance in sequence learning and parental reports of the number of words the infant used. Although this link needs to be replicated and extended, it does provide tentative support of important development of higher level supervisory attention systems involved both in learning complex sequences and in the control of language during the second year of life.

IV. Future Developments

The results that we have summarized in this chapter describe areas of the brain that appear to be active during the functions described by Norman and Shallice as requiring a degree of supervisory behavior or executive control. It is important to dissociate the components of executive control so as to provide a more analytic treatment at both the cognitive and anatomical levels. Below we consider some current efforts that might yield dissociations of the components of executive function.

A basis for dissociating conscious experience from feelings of control arise in REM sleep (Hobson, 1989). Dreaming is clearly conscious experience, but except in the rare case of so called lucid dreaming this form of conscious behavior is not accompanied by feelings of control. While we don't know the specific brain areas involved we do know that REM sleep involves the reduction or loss of the catecholamines, NE, and 5HT.

A number of recent theories have sought to determine the cognitive deficit arising from a loss of basal ganglia function. One way to characterize this deficit is to suppose that the basal ganglia are important for switching the organism between sets. Thus Parkinson patients can show either a difficulty in turning on motor behavior or a similar reduction in ability to shift task set (Keele, Davidson, Hayes & Rafal, 1995). The basal ganglia are the source of dopamine input to the anterior cingulate and thus the two frontal structures have a very close relationship. Moreover, studies of visual attention have provided rather specific hypotheses of how subcortical structure and cortical structures are coordinated in the act of orienting (Posner & Dehaene, 1994) and this could serve as a model for more anterior systems.

Lateral frontal areas appear to be involved in holding information that is not currently present in front of the mind. Thus monkey (Wilson, Scalaidhe & Goldman-Rakic, 1993 ) and human data (Jonides, Smith, et al., 1993) view lateral frontal cortex as portions of working memory. The close connection between executive function and temporary representations has played an prominent role in theories of working memory (Baddeley, 1986) and the close anatomical connections between the anterior cingulate and lateral areas of the frontal cortex may be the basis of this connection. Some places where this might occur are outlined below.

An important aspects of coherent behavior is to have a set of goals (goal tree) that can control current behavior (Carbonnell, 1981). Recently Duncan (1986,1994) has argued that goal neglect in which subjects are less able to order and implement a set of instructed goals can arise from frontal lesions. It is likely that the orbital frontal area may be of central importance to this executive function.

The analysis of executive function as a unified aspect of attention has progressed substantially both in terms of cognitive criteria and brain function. The prospect of further dissociating some of these functions and identifying them with particular structures lies largely in the future. Of course executive control raises the problem of the homunculus, but as Attneave (1959) first suggested, progress is made by dissociating operations from the overall homunculus one component at a time. As this efforts proceeds, it may be possible understand to what extent executive attention is best thought of as a unitary system, or a more distributed system.

References

Abdullaev, Y. G. & Posner, M. I.. (Submitted). Time course of activating brain areas in generating verbal associations. Science.

Allport, D. A. (1980a). Attention and performance. In G. Claxton (Ed.), Cognitive Psychology (pp. 112-153). London: Routledge & Kegan Paul.

Allport, D. A. (1980b). Patterns and actions: Cognitive mechanisms are content-specific. In G. Claxton (Ed.), Cognitive Psychology (pp. 26-64). London: Routledge & Kegan Paul.

Allport, D. A., Styles, E., & Hsieh, S. (1994). Shifting intentional set: Exploring the dynamic control of tasks. In C. Umilta & M. Moscovitch (Eds.), Attention and performance XV: Conscious and nonconscious information processing (pp. 421-452). Cambridge, MA: MIT Press.

Attneave, F. (1959). In defense of the homunculi. In W. A. Rosenblith (Ed.), Sensory Communication: Contributions to the Symposium on Principles of Sensory Communication (pp. 777-782). Cambridge, MA: MIT Press.

Baddeley, A. D. (1986). Working memory. Oxford: Clarendon Press.

Ballantine, H. T., Levy, B. S., Dagi, T. F., & Giriunas, I. B. (1977). Cingulotomy for psychiatric illness: Report of 13 years' experience. In W. H. Sweet, S. Obrador, & J. G. Martin-Rodriguez (Eds.), Neurosurgical Treatment in Psychiatry, Pain, and Epilepsy (pp. 33-353). Baltimore: University Park.

Bench, C. J., Frith, C. D., Grasby, P. M., Friston, K. J., Paulesu, E., Frackowiak, R. S. J., & Dolan, R. J. (1993). Investigations of the functional anatomy of attention using the Stroop test. Neuropsychologia, 31(9), 907-922.

Benes, F. M. (1993). Relationship of cingulate cortex to schizophrenia and other psychiatric disorders. In BA Vogt & M. Gabriel (eds.), Neurobiology of Cingulate Cortex and Limbic Thalamus. New York: Birkhauser, pp. 581-605.

Broadbent, D. E. (1958). Perception and communication. New York: Academic Press.

Broadbent, D. E. (1973). Decision and stress. New York: Academic Press.

Carbonnell, J. (1981). Subjective understanding of belief systems. Ann Arbor MI: U. Michigan Press.

Clohessy, A. B. (1994). Visual and anticipation and sequence learning in four and ten-month-old infants and adults. Unpublished doctoral dissertation. University of Oregon.

Colebatch, J. M., Cunningham, V. J., Deiber, M.-P., Frackowiak, R. S. J., & Passingham, R. E. (1990). Regional cerebral blood flow during unilateral arm and hand movements in human volunteers (Abstract No. 9P). Physiological Society.

Corbetta, M. (Chapter in this volume).

Corbetta, M., Miezin, F. M., Dobmeyer, S., Shulman, G. L., & Petersen, S. E. (1991). Selective and divided attention during visual discrimination of shape, color, and speed: Functional anatomy by positron emission tomography. Journal of Neuroscience, 11, 2383-2402.

Corkin, S., Twitchell, T. E., & Sullivan, E. V. (1979). Safety and efficacy of cingulotomy for pain and psychiatric disorders. In E. R. Hitchcock, H. T. Ballantine, & B. A. Meyerson (Eds.), Modern Concepts in Psychiatric Surgery (pp. 129-163). Amsterdam: Elsevier.

Corteen, R. S., & Dunn, D. (1974). Shock-associated words in nonattended message: A test for momentary awareness. Journal of Experimental Psychology, 102, 1143-1144.

Corteen, R. S., & Wood, B. (1972). Autonomic response for shock associated words in an unattended channel. Journal of Experimental Psychology, 94, 308-313.

Curran, T., & Keele, S. W. (1993). Attentional and nonattentional forms of sequence learning. Journal of Experimental Psychology: Learning, Memory, and Cognition, 19(1), 189-202.

Damasio, A. R. (1994). Descartes' error: Emotion, reason, and the human brain. New York: G. P. Putnam.

Dawson, M. E., & Schell, A. M. (1982). Electrodermal responses to attended and unattended significant stimuli during dichotic listening. Journal of Experimental Psychology: Human Perception & Performance, 8, 315-324.

Dehaene, S., Posner, M. I., & Tucker, D. M. (1994). Localization of a neural system for error detection and compensation. Psychological Science, 5(5), 303-305.

Deiber, M.-P., Passingham, R. E., Colebach, J. G., Friston, K. J., Nixon, P. D., & Frackowiak, R. S. J. (1991). Cortical areas and the selection of movement: a study with positron emission tomography. Experimental Brain Research, 84, 393-402.

Desimone, R., & Duncan, J. (1995). Neural mechanisms of selective attention. Annual Review of Neuroscience, 18, 193-222.

Dolan, R. J., Fletcher, P., Frith, C. D., Friston, K. J., Frackowiak, R. S. J., & Grasby, P. M. (1995). Dopaminergic modulation of impaired cognitive activation in the anterior cingulate cortex in schizophrenia. Science, 378, 180-182.

Duncan, J. (1980). The locus of interference in the perception of simultaneous stimuli. Psychological Review, 87(3), 272-300.

Duncan, J. (1986). Disorganization of behavior after frontal lobe damage. Cognitive Neuropsychology, 3; 271-290.

Duncan, J. (1995). Attention, intelligence and frontal lobes. In M. S.. Gazzaniga (ed.), The cognitive neurosciences. Cambridge: MIT Press, 721-733.

Duncan, J., Burgess, P., & Emslie, H. (1995). Fluid intelligence after frontal lobe lesion. Neuropsychologia, 33(3), 261-268.

Early, T. S., Posner, M. I.., Reiman, E. M. & Raichle, M. E. (1989a). Hyperactivity of the left striato-pallidal projection; Part I: Lower level theory. Psychiatric Developments. 2,85-108.

Early, T. S., Posner, M. I.., Reiman, E. M., & Raichle, M. E. (1989b). Hyperactivity of the left striato-pallidal projection; Part II: Phenomenology and thought disorder. Psychiatric Developments. 2, 109-121.

Eriksen, B. A. & Eriksen, C. W. (1974). Effects of noise letters upon identification of a target letter in a nonsearch task. Perception and Psychophysics. 16:143-149.

Frith, C.D.. (1992). The cognitive neuropsychology of schizophrenia. Hillsdale, NJ: LEA(1994).

Gehring, W. J., Gross, B., Coles, M. G. H., Meyer, D. E., & Donchin, E. (1993). A neural system for error detection and compensation. Psychological Science, 4, 385-390.

George, M. S., Ketter, T. A., Parekh, P. I., Rosinsky, N., Ring, H., Casey, B. J., Trimble, M. R., Horwitz, B., Herscovitch, P., & Post, R. M. (1994). Regional brain activity when selecting response despite interference: An H215O PET study of the Stroop and an emotional Stroop. Human Brain Mapping, 1(3), 194-209.

Goldman-Rakic, P. S. (1988). Topography of cognition: Parallel distributed networks in primate association cortex. Annual Review of Neuroscience, 11, 137-156.

Grafton, S. T., Hazelton, E. & Ivry, R. (1995). Functional mapping of sequence learning in normal humans. Journal of Cognitive Neuroscience, 7, 497-510.

Heinze, H. J., Mangun, G. R., Burchert, W., Hinrichs, H., Scholtz, M., Muntel, T. F., Gosel, A., Scherg, M., Johannes, S., Hundeshagen, H., Gazzaniga, M. S., & Hillyard, S. A. (1994). Combined spatial and temporal imaging of brain activity during visual selective attention in humans. Nature, 372, 543-546.

Hobson, J. A. (1989). Sleep. New York: Scientific American Library.

Janer, K. W., & Pardo, J. V. (1991). Deficits in selective attention following bilateral anterior cingulotomy. Journal of Cognitive Neuroscience, 3(3), 231-241.

Jonides, J., Smith, E. E., Koeppe, R. A., & Awh, E. (1993). Spatial working memory in humans as revealed by PET. Nature, 363, 623-625.

Keele, S. W., Davidson, M., Hayes, A. & Rafal, R. (1995). Poster presented at the 2nd annual meeting of the Cognitive Neuroscience Society, San Francisco.

Kennard, M. A. (1954). Effect of bilateral ablation of cingulate area on behavior of cats. Journal of Neurophysiology, 41, 159-169.

Kennard, M. A. (1955). The cingulate gyrus in relation to consciousness. The Journal of Nervous and Mental Disease, 121, 34-39.

Luck, S. J., Heinze, H. J., Mangun, G. R., & Hillyard, S. A. (1990). Visual event-related potentials index focused attention within bilateral stimulus arrays. II. Functional dissociations of P1 and N1 components. Electroencephalography and Clinical Neurophysiology, 75, 528-542.

MacLeod, C. M. (1991). Half a century of research on the Stroop effect: An integrative review. Psychological Bulletin, 109(2), 163-209.

Maruff, P., Currie, J., Hay, D., McArthur-Jackson, C. & Malone, V. (1995). Asymmetries in the cover orienting of visual spatial attention in schizophrenia. Neuropsychologia.

Meltzoff, A. N. (1990). Towards a developmental cognitive science. In A. Diamond (ed.), The Development and Neural Bases of Higher Cognitive Functions. (vol. 608, p 1-37). New York: New York Academy of Sciences.

Motter, B. C. (1993). Focal attention produces spatially selective processing in visual cortical areas V1, V2, and V4 in the presence of competing stimuli. Journal of Neurophysiology, 70, 909-919.

Näätänen, R. (1992). Attention and brain function. Hillsdale NJ.: LEA.

Nakagawa, A. (1991). Role of anterior and posterior attention networks in hemispheric asymmetries during lexical decisions. Journal of Cognitive Neuroscience, 3(4), 313-321.

Newell, A., & Simon, H. A. (1972). Human problem solving. Engelwood Cliffs, N. J.: Prentice-Hall.

Norman, D. A., & Shallice, T. (1980). Attention to action: Willed and automatic control of behavior (Technical Report No. 99). Center for Human Information Processing.

Norman, D. A., & Shallice, T. (1986). Attention to action: Willed and automatic control of behavior. In R. J. Davidson, G. E. Schwartz, & D. Shapiro (Eds.), Consciousness and self-regulation (pp. 1-18). New York: Plenum Press.

Ostry, D., Moray, N. & Marks (1976). Attention practice and semantic targets. Journal of Experimental Psychology: Human Perception and Performance, 2, 326-336.

Pardo, J. V., Pardo, P. J., Janer, K. W., & Raichle, M. E. (1990). The anterior cingulate cortex mediates processing selection in the Stroop attentional conflict paradigm. Proceedings of the National Academy of Science, 87, 256-259.

Pashler, H. (1993). Dual-task interference and elementary psychological mechanisms. In D. E. Meyer & S. Kornblum (Eds.), Attention and Performance XIV : Synergies in Experimental Psychology, Artificial Intelligence, and Cognitive Neuroscience (pp. 245-264). Cambridge, MA: MIT Press.

Pashler, H., & Johnston, J. C. (1989). Chronometric evidence of central postponement in temporally overlapping tasks. Quarterly Journal of Experimental Psychology, 41 (1-A), 19-45.

Perret, E. (1973). The left frontal lobe of man and the suppression of habitual responses in verbal categorical behavior. Neuropsychologia, 12, 323-330.

Petersen, S. E., Fox, P. T., Posner, M. I., Mintun, M., & Raichle, M. E. (1988). Positron emission tomography studies of the cortical anatomy of single-word processing. Nature, 331, 585-588.

Petersen, S. E., Fox, P. T., Posner, M. I., Mintun, M. A., & Raichle, M. E. (1989). Positron emission tomography studies of the processing of single words. Journal of Cognitive Neuroscience, 1, 153-170.

Posner, M. I. (1978). Chronometric explorations of mind. Hillsdale, NJ: Erlbaum.

Posner, M. I., & Dehaene, S. (1994). Attentional networks. Trends in Neuroscience, 17(2), 75-79.

Posner, M. I.., Early, T. S., Reiman, E., Pardo, P. J., & Dhawan, M. (1988). Asymmetries in hemispheric control of attention in schizophrenia. Archives of General Psychiatry, 45, 814-821.

Posner, M. I., Petersen, S. E., Fox, P. T., & Raichle, M. E. (1988). Localization of cognitive operations in the human brain. Science, 240, 1627-1631.

Posner, M. I., & Raichle, M. E. (1994). Images of mind. New York: Scientific American Library.

Posner, M. I.., Sandson, J., Dhawan, M. & Shulman, G. L. (1989). Is word recognition automatic? A cognitive anatomical approach. Journal of Cognitive Neuroscience, 1:50-60.

Posner, M. I., & Snyder, C. R. (1975). Facilitation and inhibition in the processing of signals. In P. M. A. Rabbit & S. Dornic (Eds.), Attention and Performance (pp. 669-682). New York: Academic Press.

Raichle, M. E., Fiez, J. A., Videen, T. O., MacLeod, A.-M. K., Pardo, J. V., Fox, P. T., & Petersen, S. E. (1994). Practice-related changes in human brain functional anatomy during nonmotor learning. Cerebral Cortex, 4, 8-26.

Raymond, J., Shapiro, K. L. & Arnell, K. M. (1995). Similarity and the attentional blink. Journal of Experimental Psychology: Human Perception and Performance, 21, 653-662.

Rogers, R. D. & Monsell, S. (1995). Costs of a predictable switch between simple cognitive tasks. Journal of Experimental Psychology: General, 124, 207-221.

Rugg, M. D. & Coles., M. G. H. (eds.) (1995). Electrophysiology of Mind. Oxford: Oxford University Press.

Scherg, M. & Berg, P. (1993). Brain Electrical Source Analysis. Version 2.0. NeuroScan, Inc.

Shallice, T. (1988). From neuropsychology to mental structure. New York: Cambridge University Press.

Shallice, T. (1994). Multiple levels of control processes. In C. Umilta & M. Moscovitch (Eds.), Attention and performance XV: Conscious and nonconscious information processing (pp. 395-420). Cambridge, MA: MIT Press.

Shallice, T., & Burgess, P. W. (1991a). Deficits in strategy applications following frontal lobe damage in man. Brain, 114, 727-741.

Shallice, T., & Burgess, P. W. (1991b). Higher-order cognitive impairments and frontal lobe lesions in man. In H. S. Levin, H. M. Eisenberg, & A. L. Benton (Eds.), Frontal lobe function and injury (pp. 125-138). New York: Oxford University Press.

Shallice, T., Burgess, P., Schon, F., & Baxter, D. (1989). The origins of utilization behavior. Brain, 112, 1587-1598.

Shapiro, K. L., Raymond, J. E., & Arnell, K. M. (1994). Attention to visual pattern information produces the attentional blink in rapid serial visual presentation. Journal of Experimental Psychology: Human Perception & Performance, 20(2), 357-371.

Shiffrin, R. M.., McKay, D. P. O. & Shaffer, W. O. (1977). Attending to 49 positions at once. Journal of Experimental Psychology: Human Perception and Performance, 2, 14-22.

Simpson, G. B. (1984). Lexical ambiguity and its role in models of word recognition. Psychological Bulletin, 96, 316-340.

Snyder, A., Abdullaev, Y. G., Posner, M. I., & Raichle, M. E. (1995). Scalp electrical potentials reflect regional cerebral blood flow responses during processing of written words. Proceedings of the National Academy of Science USA, 92, 1689-1693.

Spector, A. & Biederman, I. (1976). Mental set and mental shift revisited. American Journal of Psychology, 89, 669-679.

Spitzer, H. R., Desimone, R., & Moran, J. (1988). Increased attention enhances both behavioral and neuronal performance. Science, 240, 338-340.

Stroop, J. R. (1935). Studies of interference in serial verbal reactions. Journal of Experimental Psychology, 18(6), 643-662.

Taylor, S. F., Kornblum, S., Minoshima, S., Oliver, L. M., & Koeppe, R. A. (1994). Changes in medial cortical blood flow with a stimulus-response compatibility task. Neuropsychologia, 32(2), 249-255.

Treisman, A. M., & Geffen, G. (1967). Selective attention: Perception or response? Quarterly Journal of Experimental Psychology (41A), 19, 1-18.

Tucker, D. M., Liotti, M., Potts, G. F., Russell, G. S., & Posner, M. I. (1994). Spatiotemporal analysis of brain electrical fields. Human Brain Mapping, 1(2), 134-152.

Ward Jr., A. A. (1948). The anterior cingulate gyrus and personality. In Association for Research in Nervous and Mental Disease (Ed.), The frontal lobes (pp. 438-445). Baltimore: Williams & Wilkins.

Wilson, F. A. W., Scalaidhe, S. P. O., & Goldman-Rakic, P. S. (1993). Dissociation of object and spatial processing domains in the primate prefrontal cortex. Science, 260:1955-1958.

Yantis, S. & Jonides, J. (1990). Abrupt visual onsets and selective attention: voluntary versus automatic allocation. Journal of Experimental Psychology: Human Perception and Performance, 16, 121-134.

Footnote

1Oschner, K. (Personal Communication). From lesion studies of anterior cingulate conducted at Massachusetts General Hospital.