Knowing Neurons
Artificial IntelligenceMovementNeuroscience Technologies

Robots vs. Humans: Table Tennis Showdown

By Daniel Janko

Our understanding of the brain from a basic science perspective (under laboratory conditions) is gradually improving. Even though extrapolation is a powerful tool in aiding us with translating research from the lab to the real world, more research under natural conditions (playing sports, moving around in space) is needed to evaluate and refine these theories. The lack of ‘real-world’ research can be attributed to our limited ability to measure brain activity during activities that include active movement. Researchers from the University of Florida overcame this obstacle by using special high-density electroencephalography (EEG) and advanced analysis methods to record brain activity during real-world table tennis (Studnicki & Ferris, 2023).

While not an everyday activity for most of us, table tennis has been shown to challenge our cognition in various ways: (1) due to the relatively short length of the table, players are forced to react quickly to their opponent’s movements and the ball, which is moving up to 17 m/s (38 mph) (Bootsma & van Wieringen, 1990), (2) table tennis may be an effective intervention for reduction of cognitive decline (Yamasaki, 2022), and (3) elite players have been shown to have faster reactions and better visuomotor integration skills compared to nonathletes (Hülsdünker et al., 2019). Given the cognitive complexity of table tennis, understanding the underlying neural dynamics can enhance our apprehension of the neurophysiological mechanisms involved in cognitive decline, communication between visual-related and motor-related brain regions, and decision-making.  

The main objective of the study was to understand how the brain prepares for hitting the ball and how it executes that action. Additionally, the researchers wanted to study how this brain activity may differ when playing against a human compared to a robot. Studying human-robot interaction is more important than ever since the demand for collaborative robots is growing (Rodriguez-Guerra et al., 2021). All participants played against a human player and a machine in cooperative (goal: keep the ball in play for as long as possible) and competitive (goal: win a 21-point table tennis game) ways. The researchers then defined a window of interest between 500ms before and 1500ms after each contact with the ball. Windows of interest are often used to pick out the events the study is most interested in. The EEG signal during these windows of interest was then analyzed and compared between human-human and human-robot trials.  

“Interestingly, the results suggest that participants allocate more attention resources during rallies with the robot.”

The expected activation in the occipital and parietal lobes, brain areas responsible for planning and movement, was observed following the ball hits, which is a standard neurophysiological signature of movement planning and execution (Pfurtscheller et al., 1996; Van Der Werf et al., 2010). Interestingly, the results suggest that participants allocate more attention resources during rallies with the robot. This may result from the lack of bodily cues provided by the robot. The absence of these cues, which athletes often rely on to predict the direction and other properties of the incoming ball, increased uncertainty in the processing of the stimulus (incoming ball), increasing the demand on attention. Consistent with this theory, researchers observed N200 events, typically associated with uncertainty and error monitoring (Folstein & Van Petten, 2008), in response to incoming balls from the ball machine but not from humans. 

“…the absence of bodily cues does not increase the demand for attention but decreases it.”

Alternatively, the differences between the activation elicited during the human and non-human trials can be explained by predictive coding in the mirror neuron system (Kilner et al., 2007). Based on what is known about predictive coding and its role in human cognition, the differences in brain activity can be again caused by the lack of bodily cues; however, this time, the absence of bodily cues does not increase the demand for attention but decreases it. Since there is barely any variability in the way the ball machine feeds the ball, the trajectory becomes highly predictable over time. The predictability of the incoming ball’s trajectory makes the task less attention-demanding and the so-called prediction error decreases over time for the human-robot trials. 

This research uncovers important differences in the neurophysiological responses to human vs robot opponents. Two possible explanations of the findings were provided and only future research will show which one is correct. On one hand, the lack of bodily cues from the ball machine may have caused the participants to engage more attentional resources. On the other hand, the low variation in the direction and speed of balls coming from the machine could be responsible for lower prediction error, making these trials less variable and more accessible to process. The current findings have implications not only in understanding the neurophysiology of natural movements but also in the growing field of human-robot interaction research. 

~~~

Written by Daniel Janko
Illustrated by Yang-Sun Hwang
Edited by Chris Gabriel and Liza Chartampila

~~~

Become a Patron!

Appendix:

Extrapolation – An action of estimating or concluding something by assuming that existing trends will continue or a current method will remain applicable

EEG Electroencephalogram. A technique that places electrodes on the scalp to measure electrical brain activity.

Occipital lobe – Most posterior part of the brain. This lobe houses visual cortices such as the primary visual cortex and association visual areas. 

Parietal lobe – Sitting at the top of the cerebral cortex, the parietal lobe plays a role in many functions, including somatosensation, the integration of visual input, and spatial and navigational abilities. 

N200 – Taken from the electroencephalographical (EEG) terminology, N200 refers to a negative (N200) peak in activity at 200ms (N200) after stimulus onset. 

Predictive coding – a notion in cognitive neuroscience that views the brain as a predictive machine. According to this theory, instead of processing stimuli ‘from scratch’ every time, the brain constructs predictions that are then compared to the current events, and a prediction error is calculated. Based on the prediction error (difference between prior knowledge and current situation), the decision about how to react is made. 

Mirror neurons – mirror neurons are a group of neurons in the brain that mirror observed actions. When you observe someone picking up a cup of coffee, mirror neurons in the area responsible for hand movements will also be active. 

References

Bootsma, R. J., & van Wieringen, P. C. W. (1990). Timing an attacking forehand drive in table tennis. Journal of Experimental Psychology: Human Perception and Performance, 16(1), 21–29. https://doi.org/10.1037/0096-1523.16.1.21

Folstein, J. R., & Van Petten, C. (2008). Influence of cognitive control and mismatch on the N2 component of the ERP: A review. Psychophysiology, 45(1), 152–170. https://doi.org/10.1111/j.1469-8986.2007.00602.x

Hülsdünker, T., Ostermann, M., & Mierau, A. (2019). The Speed of Neural Visual Motion Perception and Processing Determines the Visuomotor Reaction Time of Young Elite Table Tennis Athletes. Frontiers in Behavioral Neuroscience, 13. https://www.frontiersin.org/articles/10.3389/fnbeh.2019.00165

Kilner, J. M., Friston, K. J., & Frith, C. D. (2007). Predictive coding: An account of the mirror neuron system. Cognitive Processing, 8(3), 159–166. https://doi.org/10.1007/s10339-007-0170-2

Pfurtscheller, G., Stancák, A., & Neuper, C. (1996). Post-movement beta synchronization. A correlate of an idling motor area? Electroencephalography and Clinical Neurophysiology, 98(4), 281–293. https://doi.org/10.1016/0013-4694(95)00258-8

Rodríguez-Guerra, D., Sorrosal, G., Cabanes, I., & Calleja, C. (2021). Human-Robot Interaction Review: Challenges and Solutions for Modern Industrial Environments. IEEE Access, 9, 108557–108578. https://doi.org/10.1109/ACCESS.2021.3099287

Studnicki, A., & Ferris, D. P. (2023). Parieto-Occipital Electrocortical Dynamics during Real-World Table Tennis. eNeuro, 10(4). https://doi.org/10.1523/ENEURO.0463-22.2023

Van Der Werf, J., Jensen, O., Fries, P., & Medendorp, W. P. (2010). Neuronal Synchronization in Human Posterior Parietal Cortex during Reach Planning. The Journal of Neuroscience, 30(4), 1402–1412. https://doi.org/10.1523/JNEUROSCI.3448-09.2010

Yamasaki, T. (2022). Benefits of Table Tennis for Brain Health Maintenance and Prevention of Dementia. Encyclopedia, 2(3), Article 3. https://doi.org/10.3390/encyclopedia2030107

Author

  • Daniel Janko

    I am a master’s student in cognitive science at the University of Potsdam, Germany. I worked as a research assistant at UAB researching emotional processing in epilepsy patients using sEEG, and also did some work on somatosensory mapping using fMRI. Currently, I am working on a project examining potential behavioral effects of the entorhinal cortex. We hope to find biases in behavior that would correspond to the functional organization of grid cells in the entorhinal cortex. In my free time, I enjoy exercising, playing video games, cooking, traveling, and exploring new cultures (mostly through food).

Daniel Janko

I am a master’s student in cognitive science at the University of Potsdam, Germany. I worked as a research assistant at UAB researching emotional processing in epilepsy patients using sEEG, and also did some work on somatosensory mapping using fMRI. Currently, I am working on a project examining potential behavioral effects of the entorhinal cortex. We hope to find biases in behavior that would correspond to the functional organization of grid cells in the entorhinal cortex. In my free time, I enjoy exercising, playing video games, cooking, traveling, and exploring new cultures (mostly through food).