National University Corporation Shizuoka University
Body region specificity in timing learning
-The brain mechanism that enables effective discrimination between fastballs and slowballs-
……
In recent years, it has become clear that our brains perform Bayesian inference. In other words, the brain learns the statistical distribution of the task target and calculates the response that has the highest probability of success. There are various statistical distributions for daily task targets (e.g. speed and course of pitching/batting), and it is necessary to learn and distinguish between them.
Makoto Miyazaki’s laboratory at our university’s Faculty of
Informatics (lead author: Keiki Matsumura, completed the Department of Informatics in March 2022) is conducting joint research with the University of Nottingham and the University of Bradford in the UK. “Singularity” was clarified.
In other words, we discovered that by responding to the distributions of two target timings (≈ fastball/slowball) using different body parts, it is possible to learn the distributions differently. Furthermore, we found that it is possible to learn to distinguish between the two distributions more quickly by associating distant body parts such as the hand/foot with two prior distributions than with close body parts such as the index finger/middle finger. I made it. Based on this “body part specificity,” it is possible to record the timing using different body parts depending on the type of pitch, such as “the foot for a fastball and the hand for a slowball.” It is also expected that it will be possible to learn at the appropriate timing. In this way, the results of this study are expected to provide foundational knowledge for proposing ways to improve sports skills and for analyzing the secret to athletes’ excellent skills.
The results will be published in the Nature Partner Journal’s (npj) science of learning journal, npj Science of Learning.
【point】
・The brain learns the statistical distribution of task targets and calculates the response that has the highest probability of success. ・In many previous studies, the state of learning was investigated by having participants experience only a single distribution.
・However, in everyday tasks, targets have diverse statistical distributions (e.g. pitching/batting speed and course)
・This study is a timing task in which different body parts are used to respond to short-time distributions (fast balls) and long-time distributions (slow balls).
We clarified that it is possible to learn these two prior distributions differently (body region specificity).
・Furthermore, rather than close body parts such as the index finger/middle finger, distant body parts such as the hand/foot are divided into two prior distributions.
By making a correspondence, it became possible to learn the difference between the two distributions more quickly.
・This result will serve as basic knowledge for proposing methods for improving sports skills and analyzing the secret to athletes’ excellent skills.
It is expected that
[Research background]
・In recent years, it has become clear that our brains perform Bayesian inference1. In other words, the brain
Learn the statistical distribution (e.g. average and variance of pitching/batting speed and course) and choose the response that has the highest success probability (hit rate)
is being calculated. Professor Makoto Miyazaki, the representative of this research, explained that Bayesian inference is being used in timing tasks.
It was revealed for the first time in the world2.
– In many previous studies, participants were asked to experience only a single distribution and their learning status was investigated. Also,
It has been reported that learning a complex distribution requires a larger number of trials.
・However, just as opponents use multiple ball speeds and courses in a ball game, in everyday tasks, the target is exposed to a variety of different ball speeds and courses.
There is a total distribution (diversity problem). In other words, in order to effectively use Bayesian estimation in daily life,
It is necessary to learn about multiple distributions.
・This study uses a research report 3 that proposes that different timers exist for different parts of the body based on analysis of rhythmic movements.
Based on prior knowledge, we predicted the existence of “body region specificity” in timing tasks. That is, two
By assigning different body parts to each of the different prior distributions and timing responses, these two things can be
I thought that it might be possible to learn different prior distributions. ・In order to verify the above hypothesis, we conducted the following psychophysical experiment.
[Image 1: https://prtimes.jp/i/96787/35/resize/d96787-35-725194ecf141860d72d2-0.png&s3=96787-35-17181bfd71c628dbefbd68f33bc836bc-3900×1765.png ]
Figure 1
Figure 1. experimental method. (a) Flow of stimulus and response in one trial. TS: Stimulus time interval, TR: Response time interval. (b) Distribution of stimulation time intervals: short-time distribution (fastball), long-time distribution (slowball). (c) Assignment of stimulation position (right/left) and stimulation time interval distribution (short/long). Here, we show an example where the short-term distribution is assigned to the stimulus on the right, and the long-term distribution is assigned to the stimulus on the left. 【Method】
・In this research, we conducted a total of seven experiments, including five main experiments and two preparatory experiments. -Eight healthy men and women (18-26 years old) participated in each experiment (total 56 people).
・Three consecutive stimuli (S1 → S2 → S3) were presented to either the left or right of the gaze point on the display [Figure 1 (a)]. trial Within, the stimulus timing (TS) of S1→S2 and S2→S3 was the same. Participants predicted the appearance of S3 from the TS of S1→S2. I expected this and responded at the same time.
・The TS of each trial was short-time distribution (fastball) [424-988 ms (average 712 ms)] and long-time distribution (fastball) [1129-1694 ms].
seconds (average 1412 milliseconds)] [Figure 1(b)].
・The short-time distribution and long-time distribution were assigned to either the right stimulus or the left stimulus [Figure 1(c)]. ・Each participant performed 640 trials of the timing task (40 trials/session ´ 16 sessions).
・The time interval from S2 to the response (response time interval, TR) was measured and analyzed.
[Image 2: https://prtimes.jp/i/96787/35/resize/d96787-35-cc695029959123ee7be3-3.png&s3=96787-35-a1129a71cc80332a3e441dfb3374d981-796×214.png ]
[Image 3: https://prtimes.jp/i/96787/35/resize/d96787-35-25abe32d7ce289520250-1.png&s3=96787-35-43e3d37cab68cfc097c189870063edfe-513×355.png ]
Figure 2
[Image 4: https://prtimes.jp/i/96787/35/resize/d96787-35-bf50bfe69d52bcc0a961-4.png&s3=96787-35-128ee9b171cfa0e1871190cbdfaeca5f-801×81.png ]
【result】
Regardless of the difference between the two distributions
(short/long), when participants responded only with the index finger of their dominant hand, they were unable to learn the difference between the two distributions [Figure 3(a)].
However, when responding with the index/middle finger of the dominant hand depending on the difference in distribution, the two
distributions could not be learned to differentiate in the first 160 trials, but were subsequently able to learn to differentiate between the two distributions [ Figure 3(b)]. In other words, the “body region specificity” hypothesis was supported.
Furthermore, by increasing the difference between the two body parts that correspond to the two prior distributions, such as hands/feet, we were able to learn and distinguish the two distributions from the first 160 trials [Figure 3(c)].
[Image 5: https://prtimes.jp/i/96787/35/resize/d96787-35-29d71062e098543d0de5-2.png&s3=96787-35-d8b2ab345f3a7daea9f2b531dbc70db2-1829×2700.png ]
Figure 3
Figure 3. Experimental result. (a) A case in which responses were made only with the index finger of the dominant hand, regardless of the difference in the distribution of stimulus time intervals (short/long) (Experiment 1 in the paper). (b) When responding with the index/middle finger of the dominant hand depending on the difference in
distribution (Experiment 2); (c) When responding with the right/left index finger and left/right foot depending on the difference in distribution (Experiment 4) ). In addition, when responding with the right/left index finger (Experiment 3) or the foot ipsilateral to the index finger of the dominant hand (Experiment 5), the two
distributions could be learned and distinguished from 1-160 trials. [Significance of research results and future prospects]
In recent years, knowledge has been accumulated showing that our brains perform Bayesian inference. The results of this research can be said to provide one of the foundations for extending these findings to human behavior in everyday environments where diverse phenomena occur, and for linking them to applications in education and medicine. Masu. In particular, this result is expected to provide foundational knowledge for proposing methods for improving sports skills and analyzing the secret to athletes’ excellent skills. For example, if you use different parts of your body to time the ball depending on the type of pitch, such as using your feet for fastballs and your hands for slowballs, you will be able to learn the timing that is
appropriate for each type of pitch. maybe.
In the future, we hope to develop this result through the following research. ▷ Verifying the effectiveness of “body part specificity” in an environment more similar to real sports using virtual reality ▷ Verification of differences in learning speed and attainment of multiple target distributions due to “body part specificity” due to differences in sports experience and skill level
▷ Verification of “body region specificity” for people with autism spectrum disorder (ASD): One of the characteristics of people with ASD is that they are not good at sports, and it has been reported that they have a deficiency in learning target distribution. 4, 5 ▷ Identification of brain regions involved in “body region
specificity”: Brain regions such as the supplementary motor area, cerebellum, and basal ganglia are involved in timing processing and also have body region reproducibility (body map). cited as a strong candidate
[Cited documents]
1. Kording & Wolpert. Nature 427, 244-247 (2004).
2. Miyazaki, Nozaki & Nakajima. J Neurophysiol 94, 395-399 (2005). 3. Ivry & Richardson. Brain Cogn 48, 117-132 (2002).
4. Karaminis, Cicchini, Neil, Cappagli, Argten-Murphy, Burr & Pellicano. Sci Rep 6, 28570 (2016)
5. Wada, Umezawa, Sano, Tajima, Kumagaya & Miyazaki. J Autism Dev Disord 53, 378-389 (2023).
[Paper information]
Magazine published:
npj Science of Learning
Publisher:
Nature Publishing Group (UK)
Paper title:
Body-part specificity for learning of multiple prior distributions in human coincidence timing
author:
Keiki Matsumura (Department of Informatics, Graduate School of Science and Technology, Shizuoka University, completed in March 2022) Neil W. Roach (University of Nottingham, UK)
James Heron (University of Bradford, UK)
Makoto Miyazaki (Department of Informatics, Faculty of Science, Shizuoka University)
More details about this release:
https://prtimes.jp/main/html/rd/p/000000035.000096787.html