Table 2.

Experimental predictions.

FORCE MATCHING PARADIGM
What is it?	This task assesses comparator processes between motor prediction and sensory outcome (69). Participants experience a force applied to their finger by a lever, and then attempt to match this force, either indirectly by using a slider with their other hand, which drives the lever, or by using their own hand directly to press the lever. Typically, in the latter case when actions are self-generated, participants tend to overcompensate by exerting larger forces. In the framework of active inference, this happens through attenuation of self-generated sensory signals during movement, which is necessary for actions to be elicited via spinal reflex arcs, even though sensory evidence indicates movement has not yet occurred (62). Actions that are not self-generated—i.e., that are similar to external events—will lack precise priors at higher levels of the motor hierarchy, and will therefore not induce sensory attenuation.
Neuropsychiatric evidence	Patients with schizophrenia (70) and functional motor disorders (71) show reduced sensory attenuation (more accurate force matching), suggesting a failure to attenuate prediction errors for actions that are self-generated, but that are perceived to be avolitional and lacking agency. This may arise through overly precise priors at intermediate levels of the motor hierarchy (68), generating actions which lack correspondingly precise priors at higher levels such as the preSMA. This entails a failure to attribute movements as self-generated, and therefore a corresponding absence of sensory attenuation.
Prediction in TS	People with TS should show more accurate force matching, due to motor predictions with high precision at intermediate levels of the motor hierarchy, without correspondingly precise priors in higher regions that are associated with attribution of actions as self-generated (such as preSMA).
SENSORY EVOKED POTENTIALS (SEPs)
What are they?	SEPs (cortical signals induced by sensory nerve stimulation at effectors) can be used as neural indices of sensory attenuation (72). SEP amplitude is normally attenuated during self-generated movement when compared to rest. This process can be modulated by dopamine (73).
Neuropsychiatric evidence	In functional motor disorder (74) and dystonia (75), there is a lack of SEP attenuation, related to failure to attribute movements as self-generated.
Prediction in TS	A reduction in this SEP attenuation due to lack of precise priors for action at highest levels of the motor hierarchy; and, secondarily, that dopaminergic therapy (73) for tics may normalise this reduction.
INTENTIONAL BINDING
What is it?	The intentional binding task is an informative paradigm for measuring perceptions of the relation between actions and outcomes (76). Participants make voluntary actions which are followed, after variable intervals, by a tone. Participants estimate the time of their action, or of the auditory tone. Typically, actions are perceived as occurring later when followed by a tone, and tones preceded by actions are perceived as occurring earlier: the action and the tone are thus “bound” together in time, suggesting the tone is perceived as a sensory outcome of the action. Intentional binding has been proposed to arise through a Bayesian system that predicts the sensory consequences of actions (64, 77).
Neuropsychiatric evidence	Patients with functional motor disorder show reduced intentional binding (78), suggesting increased precision of (intermediate-level) action priors, while patients with corticobasal degeneration causing symptoms of an “alien limb” show increased binding in their affected arm (79), suggesting decreased precision of action priors.
Prediction in TS	Patients with TS will show reduced intentional binding, given an increased precision of intermediate-level action priors.
COMPUTATIONAL PSYCHIATRY
What is it?	A general approach in which behavioural measures or modelled parameters are integrated with functional neuroanatomical data to infer the mechanisms by which activity in neural systems generates behaviour (80).
Neuropsychiatric evidence	In corticobasal degeneration, alien limb phenomena are explained by reduced precision of action priors, illustrated by abnormal intentional binding, linked to symptom severity and dysfunctional interactions between preSMA and prefrontal cortex (79).
Prediction in TS	There are numerous opportunities to gain insights from computational psychiatry approaches: one prediction would be more accurate force matching and reduced intentional binding, associated with extent of altered functional and effective connectivity between prefrontal cortex, motor preparation areas, and basal ganglia.
COMPUTATIONAL MODELLING
What is it?	Computational modelling of motor decision processes, including drift diffusion modelling, can quantify individual differences in performance of simple motor tasks (e.g., Go vs. NoGo), by parameterizing processes such as accumulation rates in favour of releasing (“go”) over withholding an action (“nogo”) (81, 82). Motor decision processes are likely to be influenced by the precision of priors and their integration in posteriors, implemented through changes in neuronal processing at specific levels within the motor hierarchy.
Prediction in TS	Parameters such as faster accumulation rates relate to the exaggerated precision of action priors in TS, and tendency to “go” over “nogo.” Combined with functional neuroimaging in a computational psychiatry approach (80) this may enable identification of specific networks and brain regions (e.g., the putamen) underpinning aberrant active inference in TS.
DYNAMIC CAUSAL MODELLING (DCM)
What is it?	The functional neuroanatomical mechanisms determining motor behaviour, including the routing of neuronal signals through CSTC circuits, can be characterised using DCM (83). For example, in the field of oculomotor control, DCM has been applied to model eye movements in terms of a balance between precision of oculomotor priors, and precision of sensory attributes of eye movement targets (84). This approach confirmed that increased sensory precision—relating to attention to target—was underpinned by increased post-synaptic gain in V1 (85). Thus, the neuroanatomical insights to hierarchical interactions that DCM provides can be interpreted in terms of quantities like precision.
Prediction in TS	DCM can quantify the influence of neuromodulators on interactions within neural hierarchies, including CSTC circuits (86), and how dopaminergic therapies modulate these (87). In TS, DCM parameters will predict quantities such as precision of action priors, and modulatory effects of monoamines.