Fig. 2.

Learning models translated into terminology similar to Van Beers (van Beers 2009). See Table 1 for terminology. The top part describes the construction of a movement within a trial. The bottom part describes learning from trial to trial. Bold red lines indicate the situation following a non-successful trial; thinner green lines indicate the situation following a successful trial. In all models, an endpoint is constructed by adding exploration and motor noise to the aim point. The actual observable behavior is the endpoint, which is then rewarded or non-rewarded. Learning always involves adjusting the aim point. This might consist of updating the aim point only following success and not following failure (Therrien16, Therrien18, Cashaback19), but it might be that the aim point is also corrected following failure (Dhawale19)