The overall efficiency of processing in the cognitive-energetic model is said to be determined by both process (computational) and state factors (such as effort, arousal and activation). Computational mechanisms of attention include four general stages: encoding, search, decision and motor organization (Sternberg, 1969). These stages of information processing are associated with experimental task variables. The processing of unbroken (physically intact) versus broken stimuli (physically degraded) is localized at encoding; the number of items to be searched in memory or on a visual display with search; deciding target present/absent with decision; and stimulus response-compatibility with motor organization. This linkage of stages with task variables enabled a search for the locus of the information processing deficit in ADHD (van der Meere, 1988; Oosterlaan, 1996; Sergeant, 1981).

Energetic factors include three pools. The first is effort (Kahneman, 1973). Effort was conceived of as the necessary energy to meet the demands of a task. Factors which effected effort were variables such as cognitive load, which could be manipulated to increase task demands. Specifically, effort was said to be required when the current state of the organism did not meet the state required to perform a task. The second pool, arousal is defined as phasic responding which is time locked to stimulus processing (Pribram & McGuiness, 1975). By contrast, tonic changes of physiological activity were thought to represent the operation of the activation pool (Pribram & McGuiness, 1975). The activation pool was identified with the basal ganglia and corpus striatum (Pribram & McGuiness, 1975).

The cognitive-energetic model includes a third level: an overriding, management or evaluation mechanism. This mechanism is associated with planning, monitoring, detection of errors and their correction. We will use this model in organizing the research which we review with respect to the inhibition deficit hypothesis of ADHD.

The terms inhibition and disinhibition have at least 12 different meanings and experimental operationalizations. Disinhibition has been conceptualized as 1) a failure to inhibit a response to "distracting", task-irrelevant stimuli; 2) a failure of appropriate occulomotor delay in looking to a remembered spatial location; 3) a failure of prepulse inhibition of the startle eyeblink; 4) a failure to inhibit a response to a misleading cue in a spatial location task; 5) a failure to inhibit to-be-ignored stimuli in a dichotic listening task; 6) a failure to inhibit an earlier appropriate, but now inappropriate, response in an S-R compatibility-incompatibility task; 7) a failure to show latent inhibition; 8) a failure to suppress inappropriate responding in Go/No-go tasks; 9) failure to suppress inappropriate responding in the stop-signal task; 10) failure to suppress inappropriate responding in the change task; 11) a failure to inhibit an experimenter-induced prepotent response when it is no longer appropriate; and 12) a failure to slow or inhibit responding as a result of the detection of an error.

These operationalizations of disinhibition are related to information processing tasks which are reviewed in sections following the review of clinical tests noted above.

For present purposes, we will restrict this presentation to considering, briefly, disinhibition in the context of information processing tasks and energetic factors.

An inability of the subject to inhibit responses to distractors (irrelevant stimuli) provides another operationalization of disinhibition. With some exceptions (Ceci & Tishman, 1984; McIntyre, Blackwell, & Denton, 1978; Radosh & Gittelman, 1981; Rosenthal & Allen, 1980; Zentall & Shaw, 1980), the majority of studies using external distractors (radio music, noise, color discrepancy, peripheral pictures) have failed to show that ADHD children are more easily distracted than normals (see for reviews, Douglas, 1983; Douglas & Peters, 1979; van der Meere & Sergeant, 1988a). In contrast to what the distractor hypothesis predicts, task performance of ADHD children may even improve in the presence of a distractor (Abikoff, Szeibel, & Courtney, 1990; Zentall, Zentall, & Barack, 1978; Zentall & Meyer, 1987). Consequently, the disinhibition hypothesis of ADHD defined in terms of inability to resist distraction has only been partially supported.

Goldstein and Blumenthal (1995) in a small sample of ADHD children found no difference in the amplitude of inhibition of the prepulse effect. They also reported that the point of maximum reduction of the startle response was reached faster in ADHD than control subjects.

If disinhibition is equated with a failure of prepulse inhibition, these two reports are inconsistent with a disinhibition explanation for ADHD. Further, the fast automatic response of the ADHD children suggests that the automatic allocation of arousal is more efficient in ADHD children than controls. Should this finding be replicated, it would argue strongly against an arousal pool explanation of the ADHD deficit.

A study by Swanson et al. (1991) found no main effect for group on RT or errors in this task. However, ADHD children exhibited slower RTs to the right visual field targets than to the left visual field targets in the non-cued and invalid cued conditions but only when the delay between the appearance of the cue and the appearance of the target was long (800 ms) but not when it was short (100 m). However, there were no visual field differences in the valid cue condition. Controls did not show any visual field differences under any of the three conditions.

The right visual field disadvantage for ADHD subjects was not replicated by Nigg, Swanson and Hinshaw (1997, who found a left rather than a right field disadvantage. However, they reported a group main effect in which the ADHD subjects had slower RTs bilaterally over all conditions.

Novak, Solanto and Abikoff (1995) reported no differences between ADHD and control children, despite the fact that the valid - invalid effect was clearly present in the task. Tomporowski, Tinsley and Hager (1994) likewise found that ADHD children performed less well than controls and college students but did not find any visual field interaction with group.

Carter et al. (1995) reported that ADHD subjects were slower than controls and that there was a validity (valid cue) effect, slower RTs to right visual field targets, and notably at the slower 800 ms interval.

Pearson, Yaffee, Loveland and Norton (1995) reported for RT no significant group main effect nor interaction with the validity effect, although errors did distinguish ADHD children from controls.

The absence of a main effect for group in three (of five) studies suggest that this task is not robust in revealing either an inhibition or attention switch deficit in ADHD. The findings of a right hemisphere disadvantage are compromised by the failure of a left hemisphere disadvantage as well.

Van der Meere and Sergeant (1988c) used a task in which children were instructed to pay attention to the top-left to bottom-right (relevant) diagonal of a display but to ignore the other (irrelevant) diagonal. In half of the trials, a target appeared on the relevant diagonal and a "yes" response was required. A "no" response was required, if the target was not presented on the relevant diagonal. Targets in incorrect loci (foils) were presented on less than ten percent of the target trials. ADHD children did not respond to foils more frequently than controls.

However, Zahn, Kruesi and Rapoport (1991) did not find a S-R incompatibility interaction in boys with "disruptive behavior disorders". Oosterlaan and Sergeant (1995) failed to replicate the interaction of S-R incompatibility with ADHD and argued that this was due to the event rate used. Indeed, van der Meere, Vreeling and Sergeant (1992) demonstrated that event rate interacted with ADHD, with ADHD children becoming slower in this task in the slow event rate. The role of event rate and its place in the energetic system will be discussed later in this chapter.

S-R incompatibility is moderately robust in differentiating ADHD children without LD from both controls and LD children without ADHD. Thus ADHD children may have a motor output deficit which may be dependent on the event rate of the task. As discussed below, this finding implicates an energetic dysfunction in ADHD.

Inhibition of prepotent responses

Latent Inhibition Latent inhibition is produced when a stimulus is passively pre-exposed for a number of trials and there is a subsequent decrease in the ability to form a new association of the pre-exposed stimulus with a new event (Lubow, 1989). The passive stimulus exposure presumably builds up an inhibitory effect in establishing a new conditioned relationship at the encoding stage. Lubow and Josman (1993) compared ADHD and control children. No latent inhibition effect was found in the ADHD group. This finding is consonant with the hypothesis that ADHD children have a weak Behavioral Inhibition System (BIS) (Gray, 1982) does not react to the "novelty" of the pre-exposed stimulus, and thus no habituation has occurred when it is presented later as conditioning stimulus. Alternatively, this result can be interpreted as suggesting that disinhibitory deficits in ADHD are not due to "stimulus" factors, that is input side effects, since in normals the inhibitory effect is caused by passive pre-exposure to stimuli. This finding is consonant with studies showing that external distractors do not have detrimental effects upon information processing of ADHD children (see for review, Douglas & Peters, 1979).

Go/No-Go task Several studies have shown that, when ADHD children are instructed to respond Go on signal trials and refrain from responding on No-Go trials, they commit more No-Go responses or errors of commission (Iaboni, Douglas, & Baker, 1995; Milich et al., 1994; Shue & Douglas, 1992). Higher proportions of No-Go responses have also been reported in children with "attentional" problems (Grünewald-Züberbier, Grünewald, Rasche, & Netz, 1978). These data support the inhibitory deficit hypothesis in ADHD. Unclear, however, is the mechanism involved in this deficit e.g., motor selection or preparation.

Stop-signal task The stop-signal task is currently the most direct measure of the processes required in inhibiting a response (Logan & Cowan, 1984; Logan, Cowan, & Davis, 1984). In this task, subjects are instructed to respond when a signal is presented and to inhibit their intended response to the signal when a stop- signal (usually a tone) is presented. Stop signals are presented at different intervals before the subject's expected response. The closer the stop signal is presented to the "point of no return", the more difficult it becomes to inhibit the response. This task, in contrast to the Go/No-Go paradigm, requires suppression of a response which is already in the process of being executed.

Recently, Oosterlaan, Logan, and Sergeant (in press) conducted a meta-analysis of eight studies employing the stop-signal task studies. For the first dependent variable, the inhibition function, consistent and robust differences were found between ADHD and control children; ADHD children were less able to inhibit inappropriate responses than controls.

ADHD children also had slower SSRTs than controls. Thus, poor response inhibition was attributed to slow inhibitory processing. For the third dependent variable, ZRFT-slope, no group differences were found; ADHD children did not differ from controls in their ability to trigger an inhibitory response nor in variability in the latency of the inhibitory process. However, Conduct Disordered (CD) children showed similar impairments to ADHD children. Therefore, the meta-analytic findings do not support the notion that response inhibition deficits are specifically related to ADHD. Rather, the results suggest that poor response inhibition characterizes children with behavior characterized as disruptive (see Chapter 1).

Another issue raised by the finding that ADHD children have slower inhibitory processes (SSRT) is whether this reflects a "generally slow" mode of information processing characteristic of ADHD children, since ADHD children also have slower RT to the target the non stop-signal trials.

The stop-signal task contains a third measure, ZRFT, which is the probability of triggering or inhibitory process. Oosterlaan et al. in their meta analysis found no difference for this measure between ADHD children and controls nor between ADHD and CD subjects or CD and control children. ZRFT is unable to explain the observed lack of inhibition to the stop-signal.

The Change task For purposes of studying controlled adaptive functioning, the ability to suppress a response and subsequently initiate an alternative response (response re-engagement) has been studied in an extension of the stop-signal task; the change task. Results of studies using this task (Schachar & Tannock, 1995; Schachar, Tannock, Marriott, & Logan, 1995; Oosterlaan & Sergeant, in press) confirm earlier reports using the stop-signal task that ADHD children are less able to inhibit a response compared to controls. Furthermore, these studies replicated the finding of slower inhibitory processing in ADHD children than controls. ADHD children were also observed in two of these studies (Schachar & Tannock, 1995; Schachar et al., 1995) to have slower response re-engagement than controls. However, Oosterlaan and Sergeant (in press) failed to replicate this finding.

In summary, the stop-signal studies indicate that ADHD children are less likely to inhibit a response than controls. However, measures of inhibition do not differentiate ADHD children from CD children so that the stop-signal findings are not specific with respect to ADHD.

Sternberg's response bias task Another means of generating prepotent responses is to bias the responding of the subject to one response rather than another. In a study by Sternberg (1969) subjects were instructed to remember a memory set and compare this with sets of items presented on a display. A target was a match between a memory set item and one of the display set items. If a match occurred, a "yes" response was required, otherwise a "no" response was required. Sternberg varied both set size and target probability, the more frequent targets bias subjects towards "yes" responses; increasing the set size led to an increase of both "yes" and "no" RTs. Subjects responded faster to the most frequent response type (yes or no) independent of the set size. This finding was interpreted to demonstrate independence of search and decision as well as a bias in the motor decision process towards faster processing of the more frequent response. This finding was replicated by Brookhuis et al. (1983).

Using the response bias task, one can test whether disinhibition is related to overly fast stimulus identification and/or in a difficulty in stopping and changing the response (van der Meere, Gunning, & Stemerdink, 1996). The baseline condition had a 50 percent target - non-target probability. In the response bias condition, target probability was 30 percent, requiring 30 percent "yes" responses and 70 percent "no" responses. Although ADHD children had slower correct "yes" and "no" responses than controls irrespective of condition, they were able to stop and change their intended response to the same extent as controls. There was no evidence of impulsive scanning or decision making. Consequently, these findings indicate that ADHD children are able to inhibit and change an intended response, even in conditions with high cognitive load. Errors were equally fast for ADHD and control children. Thus, a "fast guess" or impulsive strategy was absent in ADHD children.

Sergeant and van der Meere (1988) found that as cognitive load increased in a search task, controls slowed RTE+1 in a linear fashion. ADHD children did not perform in this manner. When processing load was low, ADHD children had a much slower RTE+1 than controls. In contrast, when processing load was high and one would expect a slow RTE+1, ADHD children were faster than controls.

Recently, RTE+1 has been shown to become slowed by MPH (Krusch et al., 1996). Drugs are considered as influencing an energetic but not computational factors in the cognitive-energetic model. Krusch et al.’s finding indicates the influence of an energetic factor on inhibitory processing.

Since both overly slow and overly fast RTE+1 latencies were observed in ADHD children, and since poor inhibition would have predicted only fast RTE+1, it would seem that these findings argue against a disinhibition hypothesis. A failure to adjust to task demands, referred to as attentional allocation (Sergeant & Scholten, 1985b), would seem to offer a better account of these findings.

To recapitulate the findings with traditional clinical measures, it is unclear whether these are pure measures of response inhibition, since a variety of processes are involved. Twelve information processing tasks have been reviewed to determine support for the disinhibition hypothesis in ADHD. Three tasks: the "Go/No-go", stop-signal and change task provide clear support for this hypothesis. The stop-signal and change tasks indicate that a slow inhibitory process underlies poor response inhibition in ADHD children, but this is not specific to this group: inhibitory performance measures do not differentiate ADHD from CD children. Four tasks provide some support for the inhibition hypothesis and five tasks do not. Few reports are available to determine whether claims of an inhibition deficit are specific only to ADHD or other externalizing disorders.

We now argue that what has been considered as evidence of an inhibition deficit in ADHD may, at least in part, be explained in terms of an energetical dysfunction. Earlier we noted that performance deficits in ADHD children could be linked to the three energetic pools and the superordinate management mechanism of the cognitive-energetic model. Two of these pools, activation and effort, are especially relevant to the inhibition hypothesis in ADHD. Activation is directly related to the motor organization (output) side of the cognitive-energetic model, which has been implicated in ADHD (van der Meere et al., 1992). Effort in this model encompasses terms such as motivation and response to contingencies, which are also thought to be disrupted in ADHD children.

Response bias (ß) has been considered a measure of impulsivity (Corkum, Schachar, & Siegel, 1996). The rationale is that ß reflects the subject’s confidence or conservativeness in responding. Conservative responding means that the subject is less likely to indicate that a stimulus is a target. Risky or impulsive responding is indicated by a low ß, when the subject requires less evidence (fast responding) to decide that a stimulus is a target. One would expect that, if impulsivity characterizes ADHD children, the majority of studies would have shown that ß differentiated ADHD children from controls. In a meta-analysis by Losier, McGrath and Klein (1996), ß did not distinguish ADHD children from controls. However, two studies showed that ADHD children had a lower ß than controls (Nuechterlein, 1983; Satterfield, Schell, & Nicholas, 1994), as would be predicted from a "fast guess" or impulsive strategy of responding. Further, van Leeuwen et al. (in press) found that ß differentiated ADHD from control subjects and ß became stricter (reflecting less disinhibition) with time-on-task in ADHD subjects. Whether ß changes with time-on-task and reflects changes in strategy or activation is an open question, since previous studies have not systematically used task variables which modify ß.

Event rate differentiates controls from ADHD children in a wide variety of tasks (Chee, Logan, Schachar, Lindsay, & Wachsmuth, 1989; Conte, Kinsbourne, Swanson, Zirk, & Samuels, 1986; Dalby, Kinsbourne, Swanson, & Sobel, 1977; van der Meere et al., 1992). In general, ADHD children have been found to perform more poorly in conditions of relatively slow event rates as compared with fast and moderate event rates.

Van der Meere, Stemerdink and Gunning (1995) compared ADHD children with and without a Tic Disorder (TD) and a normal control group, using a Go/No-Go task. Stimuli were presented at three rates: a fast presentation rate (1 s), a medium presentation rate (4 s) and a slow presentation rate (8 s). Results indicated that ADHD children with and without comorbid TD made more errors of commission (i.e., responses to the No-Go stimuli) than controls. Of interest was the finding that the inefficient task behavior of ADHD-only children was related to the event rate. These children made more errors of commission in the fast and slow conditions, but not in the medium condition, thus substantially replicating the findings of van der Meere et al. (1992). These results suggest that ADHD children's lack of response inhibition is modulated by their inability to adjust their state.

Studies using different event rates indicate that performance in ADHD children is dependent upon this variable, suggesting that the activation pool of ADHD children is in a non-optimal state, particularly, in a slow event rate condition. Results of tasks in which delay is controlled by the experimenter have, in general, been interpreted as evidence of "delay aversion" in ADHD children. The usefulness of the concept of delay aversion (Sonuga-Barke, 1996) in comparison with disinhibition is, however, not clear. If an ADHD child finds waiting aversive, a logical strategy is to shorten the period to wait by selecting a smaller reward. This would seem to be disinhibition, by most definitions, as it is inefficient performance as to obtaining the most reward.

It has been argued that it is an oversimplification to conclude that ADHD children uniquely suffer from an inhibition deficit that accounts for all of the experimental findings of impaired performance on a myriad of tasks (Barkley, 1997, see also Chapter 13). Without doubt, however, the results of a wide variety of tests and tasks can be interpreted as showing disinhibition in ADHD. However, alternative explanations for many of the findings have not yet been ruled out. We have drawn attention to the role of state or energetic factors that may generate responses that suggest an inhibitory defect. Furthermore, the role of reward, response cost and event rate influences findings to such a degree that the usefulness of "disinhibition" as an overall explanatory concept can be called into question.

A second line of difficulty with an inhibition deficit explanation of ADHD is that various tasks purported to measure "inhibition" have failed to distinguish ADHD from controls. More seriously, tasks designed to measure a specific aspect of inhibition either find no evidence of an inhibition defect or find evidence which only partially can be dovetailed into an inhibition explanation. For example, Sergeant and van der Meere (1988) demonstrated that error detection and correction can occur normally in ADHD children; only at the highest level of cognitive load do ADHD children respond without due reflection.

A third point is that tasks considered as relatively pure measures of inhibition, for example the stop-signal task, when used to discriminate between ADHD, CD and comorbid ADHD+CD children are unable to find a deficit specific to ADHD (see Oosterlaan et al., in press). This finding could be interpreted to mean that all three groups are representative of a common "disinhibitory" psychopathology (Quay, 1988, 1997). Alternatively, these groups may exhibit a common deficit on the stop signal task but which is produced by (as yet unknown) different mechanisms. Whichever interpretation is later found to be correct, it does not appear that ADHD children are the only psychopathological subgroup who have inhibitory difficulties.

It is an unresolved issue as to whether the hypothesized inhibitory deficit in ADHD reflects a central (cortical) or a peripheral (motor) deficit. For example, in the stop signal task it has been shown with ERPs and the electromyogram that two classes of failures to stop can be distinguished: "central" and "peripheral" (De Jong, Coles, & Logan, 1995). The central mechanism operates by inhibiting response activation processes in cortical motor structures to prevent central outflow of motor commands. The peripheral mechanism operates by preventing the actual execution of central motor commands by blocking transmission of such commands. This latter mechanism can inhibit responses even when the central mechanism has past the point of no return. Apparent failures of response suppression need not always reflect "impulsivity" but either deficits in central or peripheral control mechanisms. It is of interest that the central mechanism described here is similar to Gray’s (1982) suggestion that central BIS enervation leads to peripheral inhibition of ongoing motor programs.

Response inhibition is part of what has been referred to as an executive function ( We have argued that the hypothesized inhibition deficit in ADHD is dependent upon the state of the subject and the allocation of energy to the tasks at hand. We acknowledge that the cognitive-energetic model is not without its limitations. For example, specificity of the relationship of measures to pools is not, at present, satisfactory. Secondly, in animal studies it is known that there is not always a direct relationship between performance and energetics (Brener, 1987).

In terms of the cognitive-energetic model, there is inadequate activation of the actual inhibitory mechanism. It seems that effort is a less attractive explanation for the ADHD deficit, since inhibitory differences can be observed, despite reward and response cost. The evidence suggests that the activation pool is perhaps necessary for inhibition of a motor response to occur and is crucial in explaining disinhibition in ADHD.