Decision Uncertainty as a Contextual Cue for Motor Memory

Background

In a penalty shootout during a soccer match, players may face two distinctly different decision-making scenarios. For example, a player may decisively kick the ball to the right corner after seeing the goalkeeper move to the other side, or he may kick the same way without being certain of the goalkeeper’s move. It is generally believed that regardless of the initial certainty of the decision, the motor memory involved in executing the action (e.g., kicking to the right) should be the same. However, is this assumption valid? This is the core question of this study.

Existing perceptual decision models suggest that once a decision is made, the subsequent motor execution disregards any prior decision uncertainty, focusing on a single motor task. However, this study challenges this view, proposing that decision uncertainty can actually serve as a contextual cue for motor memory.

Mainstream theories in motor learning suggest that the brain flexibly forms and switches between different motor memories through context inference, primarily relying on external sensory cues that directly specify actions. However, these contextual cues do not include the decision state before the action. In reality, it seems more reasonable for people to adjust their actions based on decision uncertainty, as subjective uncertainty is related to important factors such as behavioral outcomes. Whether decision uncertainty affects motor memory and if this effect extends to new contexts are the key questions explored in this study.

Source Information

This paper is authored by Kisho Ogasa, Atsushi Yokoi, Gouki Okazawa, Morimichi Nishigaki, Masaya Hirashima, and Nobuhiro Hagura. The authors are affiliated with the National Institute of Information and Communications Technology (NICT) in Osaka, Japan; Osaka University; the Chinese Academy of Sciences; and Honda R&D Co. Ltd. The research was accepted by Nature Human Behaviour on May 13, 2024, and recently published online.

Research Process

Experiment 1: Basic Learning and Testing

The experiment was divided into two phases: the learning phase and the testing phase. Before the experiment began, participants familiarized themselves with a random dot motion decision task. Then, participants were divided into two groups: the certain decision group and the uncertain decision group. The certain decision group judged 100% coherent random dot motion, while the uncertain decision group judged 3.2% coherent motion. During the learning phase, participants performed a straight-reaching movement under the influence of a velocity-dependent curl force field. To counteract the field’s interference, participants needed to produce an appropriate lateral force (compensation force), which was measured through probe trials (channel trials).

In the testing phase, participants continued the same force field task as in the learning phase, but the probe trials included movements under six levels of uncertainty. This design allowed the observation of how motor memory formed under specific decision uncertainty conditions generalized to different uncertainty levels. These levels included 3.2%, 6.4%, 12.8%, 25.6%, 51.2%, and 100% coherence motion.

Experiment 2: Decision Uncertainty as a Contextual Cue

To directly demonstrate the role of decision uncertainty as a contextual cue, researchers designed experiments matching different strengths of force fields and decision uncertainties. The experiment was divided into three groups: main experiment 2-1 and two control experiments 2-2 and 2-3. Main experiment 2-1 tested whether participants could learn different strengths of force fields under different decision uncertainty contexts (e.g., associated with certain and uncertain decisions). Control experiment 2-2 examined the role of different visual features, while control experiment 2-3 examined the role of pre-decision time.

Experiment 3: Learning Opposite Force Fields

Experiment 3 tested whether associating different decision uncertainties could help participants learn force fields related to opposite directions, using a control group design. The experiment included two main experiments, 3-1 and 3-2, testing different strengths of force fields, and a control experiment 3-3 to further verify visual features.

Experiment 4: Transfer of Decision Uncertainty Context Between Different Visual Stimuli

Researchers examined the transfer of motor memory under the same decision uncertainty between two types of visual stimuli: random dot motion and arrow sequences, in Experiment 4. This experiment matched the subjective uncertainty levels of the two visual stimuli to determine whether transfer between different visual stimuli was possible.

Experiment 5: Decision Uncertainty in Planning and Execution Phases

In Experiment 5, researchers further explored whether decision uncertainty could serve as a contextual cue affecting motor memory in both planning and execution phases. Participants were asked to perform a straight-reaching action first and then complete a subsequent action according to the decision direction. The effects of different levels of decision uncertainty on the first phase action were reported and evaluated.

Main Research Results

Experiment 1 showed that participants successfully learned and compensated for the force field under their respective uncertainty conditions, but in the testing phase, the two groups exhibited different levels of force field compensation due to decision uncertainty.

In Experiment 2, results indicated that participants could learn different strengths of force fields based on decision uncertainty contexts. Control experiments 2-2 and 2-3 showed that this effect was not merely due to differences in visual features or pre-decision time.

Experiment 3 results demonstrated that participants could learn to compensate for opposite force fields based on different decision uncertainty contexts, confirming the conclusions of Experiment 2-1.

Experiment 4 further proved the abstract nature of decision uncertainty contexts, independent of specific stimuli. Uncertainty based on random dot motion decisions could be transferred to arrow sequence stimuli.

Experiment 5 verified that decision uncertainty could serve as a contextual cue affecting motor memory in both planning and execution phases. Even in cases where the decision action was not fully executed (nogo trials), motor memory based on uncertainty contexts could still be retrieved.

Significance and Value of the Study

The results of this study not only challenge the mainstream theories of decision-making and motor learning but also expand the role of contextual inference in motor memory. Intuitively, these studies enhance our understanding of how motor memory is influenced by contextual inference, emphasizing that it includes not only directly related motor control cues but also the uncertainty inherent in the decision process itself.

Highlights of the Study

1. Verified the role of decision uncertainty as a contextual cue for motor memory, aiding in a deeper understanding of the mechanisms of contextual inference in motor memory.
2. Demonstrated the transferability of motor memory between different visual stimuli, indicating the abstract nature of such memory.
3. Explored the influence of uncertainty in both planning and execution phases on motor memory, enriching the understanding of the neural mechanisms of the decision-making process.

This study not only provides new perspectives for research in motor learning and decision science but also may have practical applications, such as strategy optimization in sports training and rehabilitation. Future research should further explore the generality of this uncertainty context in different tasks and its neural mechanisms.