A Dual-Process Model of Response Inhibition: Insights from a Neurocognitive Perspective

Abstract

Perhaps the most critically important cognitive mechanism for survival and social cohesion is the ability to withhold an action that has been rendered maladaptive or inappropriate by altered environmental demands. There is a large body of empirical research investigating this process, which is commonly referred to as response inhibition, but which in most instances more precisely could be termed reactive inhibition because it constitutes only one element of the overall inhibition of an action. Alongside reactive inhibition, though, and certainly of at least equally import, is the capacity to recognise erroneous stimulus-response patterns in one’s own behaviour and to remediate them where they arise. This has been termed proactive inhibition and has received substantially less experimental interest until very recently, despite almost certainly contributing to overall response inhibition. Although these two cognitive mechanisms, reactive and proactive inhibition, are necessarily interdependent, they are representationally distinct and are therefore likely implemented by separate biological and cognitive processes. The basal ganglia are largely responsible for the coordination of motor control, and its neural connections to the motor and frontal cortices plan, select, and direct any intended movement, and indeed certain unintended movements also. Owing to an incomplete physiological characterisation of this circuitry until only the last decade, a critical re- evaluation of those motor functions that rely on computational cognition is germane. It is likely that reactive inhibition recruits internal basal ganglia pathways, perhaps in accordance with the classical dual-organisation model of direct and indirect pathways, because it is principally a motor function; proactive inhibition, on the other hand, requires cognitive computation, either consciously or not, and, therefore, may recruit a recently-described hyperdirect pathway that connects the basal ganglia to a prefrontal neural population that has previously been associated with overall response inhibition, but whose role has been theoretically inconsistent with motor models of inhibition because prefrontal regions are associated with higher cognitive functions and not motor function. With these limitations in mind, in this thesis, I present the experimental findings of four empirical investigations into the neurocognitive architecture of proactive inhibition using updated models in order to revise the understanding of response inhibition and, in particular, the role and underlying properties of proactive inhibition, which we operationalise as post- error slowing (PES) of reaction time. In the first study (N = 264), we investigated the role of two dopaminergic single- nucleotide polymorphisms (DRD1 rs686 and DRD2 rs1800497) which are differentially expressed along basal ganglia pathways in behavioural performance on a Go/No-Go task (the Sustained Attention to Reaction Time task, SART). We found that in those with a higher ratio of D1:D2 receptors (i.e., more rs686 A and rs1800497 T alleles) PES was engaged to a higher degree and that older age magnified this genetic effect (p < .001). In addition, we observed an interaction between age and a general factor of intelligence, g, on PES, whereby older age and lower estimates of g predicted higher recruitment of PES (p < .001). This supports the hypothesis that proactive inhibition appears to be a naturally-occurring compensatory mechanism which manifests in individuals whose reactive inhibition may be suboptimal, and indicates that the extent to which PES is engaged depends on increased dopamine D1 and decreased D2 neurotransmission. The neural generators of overall response inhibition are well described, but very little effort has been given to proactive processes. If reactive inhibition is largely motoric, then its sources can be localised using various techniques that image neural regions using haemodynamic response, but since proactive inhibition is largely cognitive, it is necessary to use other methods. To investigate the cognitive architecture of proactive inhibition we used electroencephalography (EEG). To do this, we use stimulus- and response- locked neural activity to compare the four major accounts of PES. These accounts each have wide support, explain behavioural data, and can be simulated using computational methods. We administered the SART once again to N = 100 healthy young adults and recorded their brain activity using EEG. Our results provide support for an attentional account of PES that supposes errors disturb, or disorient, attentional processing on subsequent trials indexed by the anterior N1. The N1 was significantly blunted by errors (p = .020) and the post-error N1 was correlated with magnitude of PES (p = .016). In addition, we provide additional support for our previous findings indicating an effect of age and g on PES. Here, we find that the post-error N1 diminishes with natural ageing, however, higher estimated g seemed to rescue these age-related deficits (p < .0001). These results bring into question our previous hypothesis that PES is a compensatory mechanism. Rather, it may be a consequence of disruptions to processing that incidentally improve response inhibition as a function of that disruption which offsets the initiation of response execution. Our third study was conducted to investigate the potential efficacy of neurostimulation techniques in the modulation of response inhibition and other cognitive and behavioural functions using transcranial direct current stimulation (tDCS). This study had two experiments. The first investigated whether such functions could be modulated, and the second investigated the nature of that modulation, namely, whether it could be attributed to neuroplastic induction measured by changes to motor evoked potentials using transcranial magnetic stimulation. In the first experiment, our participants (N = 56) attended three sessions, a baseline session followed the following day by single-blind, randomly allocated stimulation testing sessions separated by two days, one with a sham control, and the other with active anodal tDCS to the motor cortex. We administered a Simple and Choice Reaction Time (RT) task, the Inspection Time task, and the SART. This battery allows us to disambiguate perceptual, motor, and cognitive elements of a physical action. We observed no effect on either RT or Inspection Time and observed an effect on the proactive process on the SART (p = .002), such that PES was engaged to a smaller degree after active stimulation compared to both baseline and the sham condition. Likewise, we observed somewhat quicker RT in the SART under active stimulation (p = .073), likely because of the absence of PES, as well as more errors (p = .026), potentially indicating that PES may protect against failures of response inhibition. We attribute these results to the location of the cathode, over the right supraorbital region, roughly above the right inferior frontal gyrus. The anode in tDCS is thought to synchronise neural activity and induce long-term potentiation-like neuroplasticity, whereas the necessary cathode is thought to disrupt such synchronicity. As such, we may have disrupted prefrontal cortical functioning briefly, which in turn eroded proactive functioning. This provides reasonably strong support for frontal regions being implicated in proactive, but not necessarily reactive, inhibition, although we cannot conclude this since overall response inhibition was somewhat disrupted. The final study addresses the theoretical and conceptual limitations in existing response inhibition tasks by implementing a recent Bayesian Ψ adaptive staircase (Livesey & Livesey, 2016) in novel instantiations of two Stop-Signal Tasks (SSTs) that we developed for the purpose of directly observing behavioural proactive inhibition in two forms that are explicitly separable to the reactive process. The Ψ staircase provides an algorithm which allows for rapid estimation of SSRT in very few trials, the importance of which lies in the populations whose response inhibition and behavioural and motoric regulation are impaired due to psychopathology or neurodegeneration. Task duration is a considerable limitation on reliable estimates of performance on such tasks, and particularly in such populations. We administered four tasks (two SSTs and two Go/No-Go tasks) to N = 123 healthy young adults. We included a manipulation that cued the probability of a Stop/No-Go trial in the two SSTs and one of the Go/No-Go tasks, which was a modified form of the SART. These two probability conditions allow us to compare RT in each condition on Go trials, under the assumption that longer RT in higher p(Stop/No-Go) conditions indicates a predictive form of proactive inhibition. This is distinct from the remedial form, post-error slowing, that can still be observed in the tasks. We report two important findings. The first is that the Ψ staircase is highly successful in rapidly converging on reliable estimates of SSRT in as few as 20 stop trials, which could prove useful in designing considerably shorter tasks in the future without sacrificing reliability. Secondly, we show that predictive and remedial forms of proactive inhibition are consistently engaged in all tasks, potentially providing another avenue for thinking about proactive inhibition in the future. Thirdly, we show that estimates of SSRT, which aims to assess reactive inhibition, are robust against proactive inhibition. Taken together, the conclusions reached in this thesis represent a critical update of the neurobiology that underlies newly-discretised cognitive processes that contribute to response inhibition, as well as their psychophysiological characteristics. We have demonstrated that proactive inhibition at least partly reflects a compensatory mechanism that appears to be naturally-occurring in individuals whose reactive processes may be insufficient for psychological and biological reasons as well as individual differences in intellectual capacity. Furthermore, we present and validate a novel, theoretically cogent task paradigm to measure what we posit are discrete processes within the proactive process: remedial and predictive proactive inhibition. Given what appears to be a naturally-occurring compensatory mechanism alongside post-error slowing that corresponds to the timing of a pre-error negative inflection in electrophysiological recordings, this work raises fascinating questions about the distinction between conscious, preconscious, and subconscious brain states and their effect on behaviour.