Cognitive flexibility, OCD and the brain

This scientific commentary refers to ‘Unbalanced fronto-pallidal neurocircuit underlying set shifting in obsessive-compulsive disorder’ by Kim et al. (https://doi.org/10.1093/brain/awab483).

Cognitive flexibility is a potentially important construct for understanding brain function, with implications for clinical remediation as well as education, creativity and conceivably, political attitudes. In this issue of Brain, Kim and colleagues¹ add to evidence that patients with obsessive-compulsive disorder (OCD) may owe some of their symptoms to a fundamental problem in controlling their actions and thoughts in a goal-directed manner—associated with neural changes that result in an imbalance in striatal outflow.

The tendency to perseverate maladaptively in behaviour is of course a symptom manifest in a large number of neurological and psychiatric disorders, notably including damage to or dysfunction of the frontal lobes and basal ganglia. The neural basis of cognitive flexibility has been the subject of recent reviews² and we will consider this evidence in the light of Kim et al.’s findings, as well as the viability of the construct itself.

Cognitive flexibility can be defined as the ability to shift behaviour in response to environmental feedback, instruction or spontaneously, and it is one of a triad of functions usually considered to comprise executive function or cognitive control, the others being working memory and response inhibition. Neuropsychological tests of these functions normally employ ‘impure’ tasks that may involve all of these capacities, though perhaps to different extents.

Friedman and Miyake³ developed a method for analysing a cluster of such tests given to the same human populations, that depended on a principal component analysis for identifying variance shared by the tests but also unique to particular functions. This led to a ‘Unity and Diversity’ viewpoint that there is a common factor, general executive function, but also specific functions, such as ‘updating working memory’ and ‘cognitive flexibility’. Cognitive flexibility was the component with lowest heritability. These findings have important implications for clinical disorders, many of which are associated in part with general executive deficits, which implies a rather monolithic role that may not be helpful in the understanding of diverse symptoms. However, the ‘Unity and Diversity’ viewpoint allows both types of contribution.

Cognitive flexibility is generally identified by shared variance between tests such as the classical Wisconsin Card Sorting Test (WCST), the Trail-Making Test and other tests commonly used in experimental studies, such as Task Set Switching (TSS) (where participants are required rapidly to switch between two simple rules, entailing a ‘switch-cost’). It is still an open question as to whether cognitive flexibility as measured by these different tests can be fractionated further, for example, in response to negative feedback following a choice (as in the WCST) or simply switching in response to an external command (as in TSS) or spontaneously, as in tests of creativity.

Some suggestion of this fractionation is provided by studies in experimental animals. Dias et al.⁴ found that excitotoxic lesions of different parts of the marmoset prefrontal cortex (PFC) differentially impaired extra-dimensional set-shifting (EDS) (ventrolateral PFC) and reversal learning (orbitofrontal cortex, OFC). Analogous findings have been found for rodents and humans in neuroimaging studies. EDS is perhaps the fundamental component of the WCST, whereas reversal learning simply swaps the reward contingencies associated with two objects or stimuli. The CANTAB Intra-Extra Dimensional (IED) version of EDS (as used by Kim et al.¹) has been used in many clinical studies, including in OCD, for which a recent meta-analysis⁵ confirmed a reliable impairment (medium/large effect size) in 335 patients as compared with 311 control subjects.

There is also quite consistent evidence in humans (based on effects of regional lesions and neuroimaging) for a role for the OFC in reversal learning and for the lateral PFC in EDS. Vaghi et al.⁶ found that functional connectivity, as measured by resting state activity, between the ventrolateral PFC and caudate nucleus, was related to performance on the EDS in a group of OCD patients and controls—this finding contrasting with performance on another test of executive function (planning), which depended on anatomically distinct fronto-striatal connectivity. Chamberlain et al.⁷ found that OCD patients and their unaffected first-degree relatives exhibited underactivity in the OFC and posterior parietal cortex in a functional MRI study of reversal learning. A possible role for parietal cortex would be consistent with its contributions to attentional processes.

The results from the Kim et al.¹ study results go further than previous studies, to suggest that the EDS impairment may be related to imbalance between the direct and indirect striatal output pathways. As well as finding the usual impairment in EDS in OCD—though in a large group of drug-naïve and unmedicated patients, and in the absence of simple learning impairments—they also found that functional hyperconnectivity between the ventrolateral PFC and external globus pallidus was related to superior performance on the EDS in this group. Given the complicated anatomy and functional relationships within the basal ganglia, this could be interpreted as a functional compensatory inhibition of the subthalamic nucleus (STN), thus enabling effective attentional set-shifting. In partial support of this view, it was recently found that deep brain stimulation (DBS) of the STN in patients with severe OCD selectively improved EDS performance as well as OCD symptoms (Y-BOCS scores) with only minor effects on mood (MADRS scores).⁸ Furthermore, tractography revealed that the DBS enhanced ventrolateral PFC activity, consistent with the findings above. Incidentally, DBS of the ventral capsule also significantly improved Y-BOCS and MADRS scores without improving EDS⁸; this suggests that cognitive inflexibility can be only one facet of OCD.

Kim et al.¹ additionally used Trail-Making, which requires a form of continuous task-set switching, as a further test of cognitive flexibility in their OCD patients. Although its performance was significantly correlated with that of EDS and there was overlap in the involvement of ventrolateral PFC-pallidal connectivity, Trail-Making also engaged other fronto-striatal circuitry including the dorsolateral PFC. Reversal learning impairments presumably involve yet other fronto-striatal interactions, most likely involving the OFC.

Overall, these findings are promising, but there are many issues to be resolved. Although human studies of the EDS in different patient groups and controls are consistently finding evidence of fronto-striatal involvement, the precise regions implicated vary, for example, in the case of early Huntington’s disease.⁹ We also need to understand better the contribution of these cognitive impairments to symptoms.

It is possible that different approaches to cognitive flexibility will find closer associations with basic learning functions, which are traditionally equated by neuroscientists with processes of neuronal plasticity. The distinction between reversal learning and EDS described above was originally postulated by Mackintosh and Sutherland’s attentional theory of learning, consistent with learning the values of particular exemplars (reversal) and relating these to more hierarchical learning, based on abstract dimensions (EDS). A computational approach may further illuminate this theory; an early paper by Krushke¹⁰ did indeed find that different computational models were required for successful reversal versus EDS.

Comparisons of large groups of patients with differing diagnoses and healthy control subjects, utilizing sophisticated neuroimaging methods will ultimately resolve these important theoretical and clinical issues. Meanwhile, we may have to resort to neuromodulation, by stimulation, drugs or even cognitive training, to ‘rebalance’ neural circuits that optimize cognitive flexibility and adaptive behaviour.

Competing interests

T.W.R. discloses consultancy fees and royalties associated with CANTAB (Cambridge Cognition).