The ability to choose our actions freely is a hallmark of normal human cognition, and associated with the clear sense that we own or can control our own actions (our sense of Agency). However, the mechanisms and properties of such ‘freely chosen’ actions remain controversial. When choosing an action, we might in principle choose whether to act at all, when to act, and which action to take. All three types of decision are relevant to voluntary behaviours, but in this program, we focus on the choices between different options. We have used neuroimaging to compare ‘chosen’ and ‘specified’ actions, both in terms of the regional brain activation (ref) and the interactions within an extended motor network (ref), and then examine the impact of healthy ageing and Parkinson’s disease on the mechanisms of volition (ref).

Our current work looks beyond simple comparisons, to consider the mechanisms by which choices may be made. We use computational models to map from behavioral profiles (reaction time and response distributions) to underlying neuronal decision mechanisms. These underlying processes (inferred, latent variables) are then used to interpret neuroimaging data, using fMRI and M/EEG (ref). We also use Bayesian models to examine the internal models of voluntary actions, and the perceptual consequences of an action being considered as ‘our own’.

Having established a network or regions related to voluntary decisions, and their different roles within the decision processes, we test the criticality of regions in two ways. First, in healthy subjects, we use TMS to temporarily perturb cortical function in healthy adults. Second, we study patients with abnormalities of volition, arising from focal or neurodegenerative disease.

In a changing environment, we need to be able to adapt our behavior. For example, we need the flexibility to update our responses, and response strategies. This is required when our choices stop leading to the favorable outcomes we expect, or when an unexpected event occurs that requires change in behavior. The failure to stop an action when necessary leads to motor impulsivity. The failure to adapt our choices to new action-reward contingencies is leads to choice-impulsivity.

In this program, we examine the neural systems that regulate motor impulsivity (exemplified by the stop-signal paradigm) and choice impulsivity. By combining structural MRI (DTI), functional imaging (fMRI and MEG) with performance metrics, we address the following questions: (a) how do the systems for action selection interact with neural systems for inhibition? (ref); (b) is there concordance between structural and functional connectivity within inhibition/impulsivity networks?; (c) in neurodegenerative disorders associated with impulsivity, such as Parkinson’s disease, what are contributory mechanisms for impulsivity? And how can these systems be augmented or compensated for to improve behavioral control.

Whether a decision or action is correct or not depends on the context in which it is taken. The context draws together diverse knowledge about the likely consequences of different actions and whether those outcomes meet our current goals. The contextual knowledge might have been specified by the experimenter as a rule (e.g. if stimulus A then go left, if stimulus B then go right), or it may have been chosen by you (e.g. here is a round ball – shall I play football or basketball). The context is also affected by your motivational state (e.g. hunger, thirst, monetary interest). Often, you will be aware of this factors in advance, and can establish an ‘online’ representation of the task ‘rules’ that will help guide a quick and accurate response to events (especially where those events are to some extent predictable). This online preparation is known as ‘task set’ or ‘cognitive set’, and is frequently associated with activation of a broad fronto-parietal network. This set activity is affected by damage to the frontal lobes (ref xxx) and Parkinson’s disease (ref xxx)

Across multiple cognitive and neural systems, a general property of healthy brain function is the ability to predict forthcoming events (predictive coding). This process occurs in motor, sensory and higher cognitive systems. For example, in the motor system, with efferent copies of motor commands enabling accurate predictions of the sensory consequences of action. Within the CamCAN project, we are examining the effects of ageing on the precision of sensory prediction, and reduction in sensory perception that results from accurate prediction.

In health, the brain also rapidly identifies regularities in sensory stimuli (ref), as an automatic precursor to many cognitive processes. Higher order visual percepts also invoke predictions related to visual object identity or auditory rhythms (ref). We use a multi-mismatch paradigm to examine sensory predictive coding in dementia, using MEG, EEG and electrocorticography. Although the primary sensory cortical responses to regular stimuli are unchanged, dysfunction in the distributed neural circuits which establish sensory predictions, means that the responses to unexpected stimuli are attenuated. This provides a robust framework with which to assess diverse neurodegenerative disorders, and the study of functional neural networks even with disorders that severely impair patients’ ability to engage with complex neuropsychological paradigms.

The neural structures and systems for motor and sensory decisions differ markedly between individuals. These differences can be exploited to study neural mechanisms through structure-function correlations in the normal population. However, differences may also represent at-risk states or markers of resilience, in an ageing population. This final project therefore draws together the methods and populations outlined above to assess their potential as diagnostic markers (ie. cognitive or imaging biomarkers), or prognostic markers (of progression to dementia or cognitive decline).

Our work in this area includes collaborations with the University department of Clinical Neurosciences to predict dementia in Parkinson’s disease (ICICLE-PD study: funded PUK), and the baseline cross sectional data in healthy and cognitively impaired members of the 3000-strong CamCAN community cohort (BBSRC). In addition, new connectivity based biomarkers and predictors of neuropathology and cognitive decline in dementia are being tested, using fMRI and MEG based measures of macroscopic brain connectivity.