Lo and colleagues have recently described a recurrent network model of inhibitory control of saccadic eye movements based on neurophysiological observations in the frontal eye field (FEF) and superior colliculus (SC) of rhesus monkeys. This model emphasizes the proactive, inhibition-based, tonic neuronal activity that prevents the eye from moving in a countermanding paradigm. In this review I discuss the model with respect to existing literature that the authors did not mention, suggesting that proactive inhibitory control extends far beyond saccadic control and provides an interesting framework to interpret several attentional and movement disorders in humans.
One important aspect of cognitive control is inhibitory control, which includes the ability to refrain from reacting automatically to external events, to prevent internal impulses, or simply to cancel already planned action. Classically, response inhibition has been considered to arise essentially from bottom-up reactive processes (triggered by NoGo or Stop signals for example). Recently, the fact that inhibition can also be performed by an endogenous top-down signal has gained increasing interest (e.g., Brass and Haggard 2007; Kühn et al. 2009a,b). A critical issue now is to improve understanding of the interaction between reflexive and top-down mechanisms in inhibitory control.
In a recent study published in The Journal of Neuroscience, Lo and colleagues (2009) attempted to resolve this question by developing a recurrent network model of inhibitory control of saccadic eye movements based on neurophysiological observations in the frontal eye field (FEF) and superior colliculus (SC) of rhesus monkeys. The authors use a countermanding task requiring a subject to withhold the response to a go signal when an infrequent and delayed stop signal appears. The main advance provided by the spiking neural circuit model relies on the integration of the high tonic neuronal activity observed in neurophysiological recordings. This high baseline activity (dropping after the target has been presented) is assumed to be related to a top-down–control signal that inhibits movement-producing neurons (i.e., before inhibition is explicitly required by an external stimulus). The model convincingly simulates both behavioral (in terms of reaction time and probability of noncanceled responses in stop-signal trials) and electrophysiological (in terms of population firing rates of movement and fixation neurons) data. It provides new insight into the meaning of inhibition-based tonic neuronal activity and sheds light on its top-down proactive nature. It offers, for the first time, a biophysically realistic model of the circuit mechanism of inhibitory control of saccades that accounts for the interaction of top-down and bottom-up influences.
Lo and colleagues (2009) assert that the persistent proactive top-down inhibitory control is a novel hypothesis and a unique feature in their model. Although integrating this process in a recurrent network model of inhibitory control is undoubtedly innovative and promising, it seems that this advance would benefit from being discussed in the context of work that the authors did not mention. Indeed, it appears that the concept of proactive top-down inhibitory control has already been developed over the past 3 years by Boulinguez and colleagues in behavioral and neuroimaging studies in humans (Boulinguez et al. 2008, 2009; Jaffard et al. 2007, 2008; Fig. 1). This mechanism was studied after the authors observed that the presentation of a warning cue in standard visual target-detection tasks elicits transient automatic activations over the sensorimotor cortex (Jaffard et al. 2007) and in the form of subthreshold activations in the muscles involved in the response (Boulinguez et al. 2008). These authors suggested that only tonic proactive inhibition over triggering processes allows preventing undesired movements. Although the observation that automatic motor activations are elicited by irrelevant visual stimuli is not new, the idea that higher-order processes of inhibition are necessary to prevent undesired automatic responses is still controversial (e.g., Aron 2007; Sumner and Husain 2008). In their series of papers, Boulinguez and colleagues hypothesized that tonic top-down proactive inhibitory control is involved as soon as there is uncertainty in the task—i.e., when unpredictable or multiple stimuli may be presented. Using functional magnetic resonance imaging (fMRI), the source of this control was mainly localized within the medial prefrontal cortex, the posterior cingulate cortex/precuneus, and the inferior parietal cortex, whereas the primary motor cortex, the supplementary motor cortex, and the putamen were identified as targeted sites of response inhibition when human subjects were involved in key-press tasks (Jaffard et al. 2007, 2008). Further event-related potential studies by the same group (Boulinguez et al. 2009) and unrelated electroencephalographic investigations (Delorme et al. 2007) in humans support the model and are consistent with the special role played by the medial prefrontal cortex in volitional inhibition (Brass and Haggard 2007; Kühn et al. 2009b).
As stated both by Boulinguez et al. and Lo et al. (2009), a better understanding of proactive inhibitory control is likely to lead to important advances in the cognitive neuroscience of attention and motor control, as well as in clinical research. In the following text I develop why these two complementary approaches would certainly benefit from referring to each other and also to other important recent studies.
The initial concept of proactive inhibitory control first introduced by Boulinguez and colleagues (Jaffard et al. 2007) clearly stated that a central aspect that is often neglected in standard methods is the activity taking place before any stimulus is presented. The authors claimed that numerous behavioral and imaging studies are probably biased because an inappropriate baseline is classically used to perform standard contrasts. Unfortunately, as they acknowledged later (Jaffard et al. 2008), fMRI studies in humans still face a strong limitation when an appropriate baseline is used because the ambiguity related to the physiological correlates of inhibition does not allow unequivocal interpretation of the blood oxygen level–dependent signal in these tasks. Fortunately, however, electrophysiological recordings in animals provide the opportunity to disentangle the activity of inhibition-related and movement-related neurons. In this respect, neuroimaging studies in humans provide important clues that may serve as a starting point for electrophysiological experiments in animals and neural modeling. Reciprocally, electrophysiological recordings provide strong interpretations of the nature of rough cerebral activation patterns observed with neuroimaging in humans. Accordingly, future research would benefit from a thorough review of the existing literature on proactive inhibitory control and related topics to build a generalized model and elucidate the entire circuitry of this top-down control.
Indeed, it is tempting to speculate that the convincing spiking neural circuit model of the top-down control of saccades developed by Lo et al. (2009) illustrates a generalized mechanism of proactive inhibitory control that is not restricted to eye movements. Although the authors raised this possibility in their report, they speculated that the dorsolateral prefrontal cortex, the supplementary eye field, and/or the anterior cingulate cortex may constitute possible origins of the top-down modulation of FEF and SC activity. Important electrophysiological data recently published in the Journal of Neurophysiology (Narayanan and Laubach 2009) support the view that the medial prefrontal cortex is a better candidate for this function, bridging the gap between human and animal studies. In that study, the authors describe firing patterns of single neurons in the dorsomedial prefrontal and motor cortex of rodents during a simple reaction time task. They provide results consistent with the prestimulus tonic activity observed by Lo et al. (2009) and clearly interpret these data in terms of proactive inhibitory control of action. It is obvious that this piece of work may serve as a basis for extending Lo and colleagues' recurrent network model in several ways: first, by considering whether proactive inhibitory control is not only involved in countermanding actions (i.e., when the probability to inhibit an already initiated action is explicit) but also generalizes to most sensorimotor interfaces and, second, by including in the model the most probable source of this top-down inhibitory control. This perspective represents a critical step in our understanding of top-down and bottom-up interactions in cognitive control since it may allow the source of the control (e.g., medial prefrontal or dorsolateral prefrontal cortex) to be disentangled from the targeted sites of inhibition (e.g., motor/premotor cortex for limb movements vs. frontal eye field/superior colliculus for eye movements) and brain areas involved in concomitant processes (e.g., conflict monitoring, attention, and response selection).
Finally, I will elaborate on the suggestion of Lo et al. (2009) that their model may have significant implications for exploring the basis of impulsivity associated with psychiatric and neurological conditions. This idea was also proposed by Boulinguez et al. (2009) on the basis of growing evidence that a dysfunction of the medial prefrontal cortex was observed in various pathologies such as schizophrenia, attention-deficit hyperactivity disorder, or Parkinson's disease (PD) and was often associated with impulsivity- or akinesia-related symptoms. However, the first real attempt to relate such behavioral deficits to a dysfunction of the network supporting proactive inhibitory control was provided just recently by a study investigating akinesia and subthalamic nucleus deep brain stimulation (STN-DBS)–induced impulsivity in PD patients (Ballanger et al. 2009). This study measured regional cerebral blood flow during a Go/NoGo task to study the motor improvement and response-inhibition deficits associated with STN-DBS in these patients. Results show that the switch from akinesia to impulsivity involves changes in synaptic activity in part of the cortical network responsible for proactive inhibition consistent with diminished control. Obviously, further clinical work is needed to extend this conclusion to other diseases and other functions (for example, to relate more cognitive aspects of inhibition to psychiatric conditions such as schizophrenia). Considering simultaneously behavioral, electrophysiological, modeling, neuroimaging, and clinical studies in both animals and humans will undoubtedly help in reaching that goal faster.
- Copyright © 2009 the American Physiological Society