Abstract
We are required to make decisions at every step of our lives. Themajority of these are made automatically, without any conscious deliberation, that is, they are implicit or associative decisions. Others are analytic, explicit or rule-based, and are made with deliberation, and perhaps some hesitation [1]. The essence of a decision is the culmination of a process in a conclusion or resolution. Sometimes we are unable to decide, and may postpone the decision, let someone else make it for us, or ‘decide’ not to make a choice at all. Decisions of course have consequences, and a decision is a good one if it maximizes desirable or minimizes undesirable outcomes [2]. Most decisions are made in a context that involves others, and therefore are influenced by social role and personal identity. Some decisions are inherently difficult to make, while others are difficult for some individuals in certain contexts. Decisions are embroiled with emotions in the evaluation of both experienced and expected outcomes. Whatever the complexity of decision-making, it is fundamental to cognition, and habitual indecision can be disabling.
Compulsive individuals are habitually indecisive. They may take hours to select a gift, which may eventually be packaged with a ‘belated’ birthday card, or be unable to decide when an assignment or project is complete, resulting in its delayed submission. Such indecision reaches its pathological apex in obsessive–compulsive disorder (OCD), which is defined by the presence of obsessions and/or compulsions. The hallmark of many of these symptoms is the presence of a doubt regarding whether a particular act has been satisfactorily concluded, or a particular event may or may not occur, or even whether the consequences of action or inaction may or may not be negative. For instance, ‘washers’ examine their washed hands unable to decide whether they are clean and free of germs. Rather than make an immediate choice and face the consequences, they wash them repeatedly, despite being acutely aware of the fact that they are actually clean, and that further washing is absolutely unnecessary. Similarly, ‘checkers’ cannot decide whether ‘the lights have been switched off’, or ‘the door has been securely locked’. ‘Hoarders’ are unable to decide whether discarding an old newspaper may not lead to regret later. For some unknown reason, the knowledge or sensory-perceptual evidence that adequately allows most individuals to reach a quick decision is simply ineffectual in patients with OCD.
Therefore, in attempting to answer the question, ‘Why is it that a patient with OCD is unable to decide not to check or count orwash or to terminate rumination over the pros and cons of an idea or situation?’ it may help to conceptualize OCD as a disorder of decision-making. Neurobiologists are beginning to understand the neuroscientific basis of decision-making in animals and humans, and this knowledge can perhaps be usefully applied to OCD for its neurobiological investigation, cognitive appraisal and behavioural treatment.
Neurobiology of decision-making
The cognitive neuroscience of decision-making is currently receiving a lot of attention with studies adopting a number of approaches to its many facets [3]. Anatomically, the brain region essential for the neural processes of decision-making is the prefrontal cortex (PFC). The evidence for this originates from lesion studies, in particular the landmark study by Harlow [4] of Phineas Gage, a railway worker who after a brutal injury to his PFC with a tamping iron engaged repeatedly in behaviours with negative consequences. Recent neuroimaging studies corroborate the role of the PFC in a variety of decision-making tasks and its subprocesses such as planning, inductive reasoning, reward processing and manipulating complex information. These studies have identified subregions in the PFC, that is, the dorsolateral prefrontal cortex (DLPFC), the orbitofrontal cortex (OFC) and the anterior cingulate cortex (ACC). The regions are not autonomous, but are engaged in multiregion neural subsystems that are separable in functions and interact with each other [5].
Decision-making requires the online retention of information, its manipulation with respect to options and the maintenance of goals. Working memory therefore plays a fundamental role [6] and the DLPFC is the major cortical region subserving this function [7]. The DLPFC is also involved in reasoning and the categorization of novel stimuli [8]. It has therefore been suggested that the DLPFC participates in decision-making tasks involving the manipulation and integration of relational information from multiple sources, especially if these processes involve working memory [9]. There may also be lateralization of functions, with poorly structured and relatively unconstrained decisions engaging the right DLPFC, and simpler decisions, or those that are well structured in terms of goals and options, engaging the left DLPFC [10].
The OFC, located on the roof of the orbit, occupies the ventral region of the PFC and is subdivided by some into ventromedial and ventrolateral portions [11]. The ventromedial OFC has been implicated in OCD using Bechara's gambling task [12], [13] and appears critical in integrating affective information relayed from other limbic areas, or signalling some types of reward directly. With affectrelated information accessible, it is likely that the DLPFC then consciously deliberates various options and compares competing objectives. At this point, information from the OFC and perhaps other limbic areas may help to emotionally bias particular options either positively or negatively [14–16]. This would help facilitate decisions when faced with complex or plentiful options by providing information that favours appropriate choices. Those choices that are rewarding would be enhanced and the ones associated with punishment or negative contingencies would be inhibited. In this context, it is interesting to entertain the possibility that ‘gut-feeling’ relates to OFC activity. When a formerly rewarding option that is no longer rewarding is later presented, the OFC is likely to be involved in suppressing its selection [16]. In this manner, the OFC maintains a representation of reward history and modifies former reward representations that may have currently become disadvantageous. In very simple, well-structured decisions, the DLPFC would require little influence from the OFC and limbic regions, as such decisions would probably not involve much emotional processing. Conversely, decisions that rely predominantly upon prior reward history probably engage the OFC quite actively with little DLPFC involvement. This is supported in part by the fact that OFC lesions have classically been associated with abnormal decision-making in daily life even though the individual retains the ability to make the correct choices [17], [18].
It is likely that the ACC, a heterogeneous region that can be partitioned architecturally and functionally [19], [20], is involved in the general monitoring of decision outcomes. This region of the PFC is most active when a decision involves conflict, that is, multiple/diverse outcomes are possible, and it therefore perhaps biases activity in other regions toward one choice or another. In the event of a poor decision outcome, the ACC may signal that changes will be necessary to avoid future repetition of such judgement errors, the ‘error detection’ role of the ACC [21]. The ACC may therefore be important for closure in situations of uncertainty or conflict, and for rapid adjustment of behavioural response.
Two other regions important in decision-making, by virtue of their connectivity with the PFC, are the basal ganglia and the amygdala. The basal ganglia are important for movement control and implicit learning, and their involvement in decision-making may include striatal dopaminergic influences on when to adjust behaviour [22]. This would be most critical after the outcome of a decision is signalled, perhaps by the ACC, necessitating reward-related behavioural modification. Lower dopamine levels are likely to increase deliberation time, as happens in Parkinson's disease and many basal ganglia disorders [23]. The amygdala receives substantive visual and auditory information and is intimately involved in emotional processing. Its extensive links to the OFC identify it as an important source for emotional input. The OFC and amygdala interact in reward processing [24], and amygdala lesions may have consequences similar to OFC lesions [12].
The PFC is networked with subcortical structures by means of well-defined parallel frontal–subcortical loops that link up specifically with the basal ganglia and thalamus, forming circuits that ultimately feedback to the PFC [25]. Three such circuits are important for highlevel decision-making [26]: the dorsolateral prefrontal circuit that is responsible for mediating executive functions and working memory; the orbitofrontal circuit that is important for the selection of appropriate social behaviour, and therefore is involved in behavioural control; and the medial frontal circuit that regulates motivation and maintains activity.
Neurobiology of obsessive–compulsive disorder and compulsive behaviour
Much of the neurobiology of OCD as it emerges appears to converge with our understanding of decisionmaking and provides the basis for our model. The majority of the evidence comes from the neurological correlates of OCD, neuroimaging studies and the psychosurgery literature.
The early neurological literature emphasized the association of OC behaviour with disorders of the basal ganglia [27]. Obsessions and compulsions were common in patients who had suffered from encephalitis lethargica, and these were often accompanied by parkinsonism, oculogyric crises and anxiety. On post-mortem examination, these patients often showed abnormalities in the basal ganglia [28]. The OC phenomena are also associated with Tourette's syndrome, Syndenham's chorea, carbon monoxide poisoning and other disorders in which basal ganglia are known to be involved, leading to the proposal of a neuroethological model of OCD with a primary emphasis on the basal ganglia [29]. However, this literature is anecdotal, and it is by no means certain that the pathology in these disorders is restricted entirely to the striatum.
More recent evidence has implicated the frontal lobes, and in particular the OFC and ACC. Neuropsychological investigations of OCD have implicated frontal lobe dysfunction in some but not all studies [30], and a study using evoked potentials provided evidence for frontal cortical abnormalities [31]. A large number of functional neuroimaging studies have provided evidence for abnormalities in specific regions of the PFC. In contrast, structural neuroimaging studies have mostly been negative, identifying no specific structural abnormality in OCD patients, with the exception of magnetic resonance imaging (MRI) studies that have revealed reduced frontal lobe white matter [32–34]. Functional studies examining cerebral metabolism using single photon emission computerized tomography (SPECT) [35] and positron emission tomography (PET) have found abnormalities in the OFC, ACC and the caudate nuclei [36–39], with the most consistent finding being an increase in metabolism in the OFC implicating specifically the right anterolateral region [40]. Increases in ACC and the caudate are reported less consistently. Studies have correlated OFC metabolic activity with symptom severity, noting that patients with more severe abnormalities have a poorer response to treatment [38], and that successful treatment with drugs or behaviour therapy reduces abnormally increased activity [41], [42]. In a PET study in which patients with OCD were stimulated in the scanner to induce obsessional symptoms, there was a significant increase in activity in the OFC, thalamus and cingulate cortex [43]. In a similar study using functional MRI, Breiter et al. [44], when comparing OCD patients with healthy controls, reported OFC, DLPFC, ACC, anterior temporal cortex, insula, striatum and the amygdala activation but only in patients.
Further evidence for the involvement of the OFC, ACC and the frontal–subcortical circuits in OCD comes from reports of improvement following various types of neurosurgical intervention, including anterior cingulotomy, limbic leucotomy, subcaudate tractotomy and anterior capsulotomy [45], [46]. These operations involve various degrees of disconnection of the OFC and/or ACC from the striatum, anterior thalamus, amygdala and other subcortical structures creating perhaps a degree of insensitivity to negative information [47]. However, the precise lesions necessary for successful remission of OCD symptoms are not known [30], although it is interesting to note that in anterior cingulotomy, the ACC may also be partially ablated [48].
The above findings have prompted a number of neurobiological models of OCD to be proposed [29],[49–51]. Common to all of these is the idea of an abnormality in the OFC with increased activity in the ACC and the striatum also being of importance. Our model of decision-making abnormalities in OC behaviour and OCD draws on the same neuroimaging and neurosurgical data. Whereas previous literature on decision-making abnormalities in neuropsychiatric disorders has focused on lesions in the OFC and ACC [52], we draw attention to the increased activity in these regions in the resting state of OCD, with further activation during symptom provocation. Moreover, during OC symptom induction, temporal lobe structures are activated. Of particular relevance is the amygdala in which activation occurs bilaterally emphasizing its importance in emotional valence, and suggesting its role in determining the content of OC symptoms perhaps by virtue of its involvement in emotional memories [53]. The normal activity of the DLPFC at baseline, and its activation during symptom provocation, suggests its role in executive decision that terminates the compulsion or the obsession.
The neurotransmitters involved in the above frontal– subcortical circuits are glutamate, gamma-aminobutyric acid (GABA) and dopamine [26]. Serotonin (5-HT) has received the most attention inOCD, because the enhancement of 5-HT neurotransmission has therapeutic effects. Indeed, potent 5-HT re-uptake inhibitors are effective treatments for OCD [54] and 5-HT2A/2C agonists may acutely decrease OCD symptoms [55]. However, 5-HT depletion does not exacerbate OCD symptoms in people with untreated OCD or fluvoxamine-treated OCD, but it does partially reverse the antidepressant responses in depressed people with OCD [56]. This argues against 5-HT dysfunction being the primary source of OC symptoms, indicating instead that 5-HT may play a modulatory role. For instance, the long-term blockade of the serotonin transporter, a condition necessary for therapeutic effect, has a number of effects. These include changes in the ratio of dopamine to serotonin turnover, reduced sensitivity of various subtypes of presynaptic 5-HT receptors and alterations in the gene expression of target neurons for disease-related stress neuropeptides [57].
Proposed model
We summarize our model of decision-making in OCD and compulsive individuals as follows.
In simple decision-making in healthy individuals, the DLPFC cortex is activated as the substratum of working memory and the region for deliberation of actions, with input from the OFC to assist with affective choices. The ACCmonitors the determination of a choice, and the basal ganglia are activated depending upon whether a rewardrelated behavioural change is involved. If the task is well rehearsed, choice determination may become automated, with the left DLPFC being preferentially involved and the role of the basal ganglia increased.
In OCD and compulsive individuals, some decisionmaking acquires an abnormal emotional valence, because of abnormal activity in the OFC, leading to prolonged deliberation before the decision is made, or repetition because of reward uncertainty. The abnormality is particularly likely to be in the lateral OFC which is sensitive to negative rewards. Executive decision is still possible in the DLPFC, but this is associated with the experience of a negative affect. The ACC that monitors this activity is abnormally activated in the process, and prolonged activity in this region delays closure of decision-making.
It is recognized that individuals with OCD have both obsessions and compulsions. The repetitiveness of an obsession must be terminated by amechanism similar to the closure of a deliberated action. The emotional valence of an obsession is again explained by the abnormal activity in the OFC. It is possible that patients with a purely obsessional illness have a more restricted brain abnormality, without increased activity in the striatum, and the extent of the abnormality in the limbic forebrain may determine the clinical features of OCD. The presence of tics in some patients suggests particular abnormality of the striatum.
The content of the obsessive–compulsive activity is determined to a great extent by the reward history. There are many clinical examples of this. For instance, a woman prone to compulsive behaviour developed compulsions after the birth of her first child, as she became acutely concerned with hygiene following an episode of diarrhoea in the baby. A man began to compulsively check for dogs and wash off saliva after a stray dog bit his child and he had to live through the anguish of possible rabies. Negative contingencies relating to weight and self-image underpin the compulsive behaviours of some anorexics. Of course, such evidence cannot be found in every case, and many compulsions are not explained by the historical experiences of an individual. It is possible that a baseline biological abnormality, such as that in the OFC, sensitizes some individuals to a particular theme, especially one that necessitates deliberation. By association, other themes and behaviours may then be encompassed. The amygdala may be important in this process, as the themes in OCD usually relate to issues that generate strong emotion in our daily lives such as dirt, germs, sex and religion.
Anxiety or arousal leads to increased activation of the OFC, possibly through the amygdala, further decompensating decision-making in OCD and increasing the aversive nature of making choices. Thus, OCD individuals are trapped in a vicious cycle of increasing anxiety and OC symptoms. In contrast, healthy patients do not show the same degree of increased OFC activity with anxiety, and their decision-making is not significantly disrupted until anxiety increases beyond a certain threshold.
In depression, activity in the DLPFC is reduced, thereby making executive decisions more difficult. As a consequence, the OFC becomes more salient in decisionmaking in depressed individuals, and sometimes compulsive symptoms may develop for the first time, or become exacerbated. The OCD patient, with a baseline of abnormal activity in the OFC, decompensates further during depression.
The OFC maintains a record of the reward history and, in the presence of negative feedback, makes future choices more affect-laden. In this context, the ACC monitors conflict and signals a need for change in strategy although this may not be practicable.
Abnormal activity in the basal ganglia may lead to the activation of automated behaviours without deliberation. This may be accompanied by reduced feedback through the subcortical–cortical circuits, thereby leading to an urge to repeat the behaviour. It is not surprising that the repetitive behaviours found in basal ganglia disorders are often not associated with the emotional valence that accompanies the compulsion of OCD, and the themes are different [58], [59]. The situation in disorders such as Tourette's syndrome may vary depending upon whether the abnormality is circumscribed or widespread [58].
Consequences of this model
The neuroscience of decision-making has seen a major resurgence and it is likely that as further developments occur, it will open up new avenues for the study of OCD. In this context, neuroimaging paradigms used for decisionmaking may be usefully applied to OCD and related disorders. The above model helps in the understanding of the extant findings in the neuroimaging literature and at the same time creates testable hypotheses. Some examples are as follows:
The model supports the use of neurosurgical interventions for the treatment of OCD, targeting the connections between the OFC and the striatum, thalamus and amygdala. It would suggest that orbitofrontal leucotomy that severs connections of both medial and lateral OFC is likely to produce the best results. Anterior cingulotomy exerts its benefit by reducing the activity of the ACC and severing frontal–subcortical fibres, whereas anterior capsulotomy achieves the same by partially severing thalamcocortical connections to the OFC and the ACC. Deep brain stimulation, which has recently been used to treat OCD [60], can similarly target these key regions of the decision-making process. With brain stimulation, the attempt is not only to temporarily disrupt a brain region but also to modulate brain activity in certain regions in order to control the symptoms of OCD. Although anterior capsular stimulation has been used with this technique [60], it should be possible to modify activity in the medial and lateral OFC using implanted electrodes.
The model argues that the activation of the DLPFC produces executive control and leads to the termination of OC symptoms. Increasing the activity of the DLPFC with repetitive transcranial magnetic stimulation (rTMS) should therefore lead to improvement in the symptoms of OCD, as has been shown in two studies [61], [62] and warrants further exploration. Theoretically, slow rTMS could be used to reduce the activity in OFC, but this is not practical because current technology does not enable direct access of a deep brain structure to TMS. However, vagus nerve stimulation is known to modulate the responsiveness of forebrain neurons and may prove to be useful in OCD and related disorders [63].
The conceptualization of OCD as a decision-making disorder suggests new possibilities for the cognitive behaviour therapy of this disorder. A behavioural analysis of OCD should involve the determination of the emotional influences on decision-making. Strategies should then be developed for prompt decision-making in various contexts, with particular situations being preferentially targeted.
Our model also encourages the exploration of neurotransmitter abnormalities other than those conventionally associated with OCD. For instance, there is empirical evidence of the influence of antidopaminergic drugs on OC symptoms, perhaps by means of striatal action. Modulation of the system and its components may be possible through the action of alternate neurotransmitters such as glutamate, GABA and the neuropeptides somatostatin and cholecystokinin. The role and actions of these should be explored.
Finally, reformulating OCD in the context of decisionmaking allows the application of novel cognitive neuroscience techniques to test symptom provocation and their abatement in order to determine further the neural substrates of obsessions and compulsions. Obsessive– compulsive disorder is one of the more precisely defined psychiatric disorders and therefore holds great promise for the advancement of psychiatric neuroscience.
Footnotes
Acknowledgements
Angie Russell prepared the manuscript. We have no financial involvement or affiliations with any organization whose financial interests may be affected by material in the manuscript.