Table 1.

Terminology

			Model
Time
$t$	Trial		All
Sources of variability
$m$, ${\sigma }_m^2$	Motor noise	Inevitable variability that is always present. Also called unregulated variability (Dhawale et al. 2019)	Cashaback19
			Dhawale19
			Therrien16
			Therrien18
$\eta$, ${\sigma }_{\eta }^2$	Exploration	Variability that can be added and can be learnt from. Also called regulated variability (Dhawale et al. 2019)	Cashaback19
			Dhawale19
			Therrien16
			Therrien18
${\sigma }_{\eta \ast }^2$		Input exploration
${\sigma }_{\eta -}^2$		Exploration estimated following non-successful trials
${\sigma }_{\eta +}^2$		Exploration estimated following successful trials
$\widehat{{\sigma }_{\eta }^2}$		(A)TTC estimate of exploration
Movement generation
$\mathit{AP}$	Aim point	Mean of the probability density of movement endpoints given a certain ideal motor command (van Beers 2009)	All
$\mathit{EP}$	End point	Observable movement outcome	All
Reward-based motor learning
$R$	Reward presence or absence	R = 0: no reward	All
		R = 1: reward
$\mathit{RPE}$	Reward prediction error	Difference between actual reward obtained and predicted reward	Dhawale19
		RPE > 0: Reward obtained
		RPE < 0: No reward obtained
$\overline{R_{\tau }}$	Low-pass filtered reward history	Low-pass filtered reward history of the τ previous trials	Dhawale19
$\alpha$	Reward-based learning parameter	Learning gain, adjustment fraction	Cashaback19
			Dhawale19
			Therrien16
			Therrien18
$\beta$	Reward rate update fraction	Gain of updating the reward rate estimate ($\overline{R_{\tau }}$) with the most recent trial outcome	Dhawale19
$\tau$[tau]	Number of trials in reward history memory window	Inferred memory window for reinforcement on past trials, or the time-scale of the experimentally observed decay of the effect of single-trial outcomes on variability (Dhawale et al. 2019)	Dhawale19