Neural feedforward and feedback mechanisms of meaningful game play

Adam Noah, Shaw Bronner and Atsumichi Tachibana

Abstract

Gamers employ a wide variety of sensory modalities to hone skills when learning to play games. Auditory, visual, and tactile cues may provide players feedback necessary to perfect fine motor performance necessary to achieve higher skills. Feedforward cuing also likely plays an important role in training of motor skills. Examples of feedforward mechanisms are: familiarity with physics of first person shooters (predicting differences in trajectories of rocket launchers versus sniper rifles), or working knowledge of the particular turns within an individual circuit on a race simulator. Rhythmic games like Rock Band and Dance Dance Revolution (DDR) integrate visual, auditory, and tactile feedback with feedforward information of the beat of the song into the design of game play. We are particularly interested in how the balance of feedback and feedforward information can influence meaningful play with respect to neural mechanisms of motor learning and game skill development. Our hypothesis was that rhythmic games provide players enhanced knowledge of timing of when to press buttons in a game as a result of feedforward information from the song rhythm. We hypothesized that players will perform better, recruiting modified neural networks, when playing with rhythm versus no rhythm.

Fifteen healthy, young subjects played a modified version of DDR while undergoing a functional MRI (fMRI) scan. Gameplay was modified so that there were only left and right arrows within "dance" sequences, and a compatible two button foot pad was built to minimize motion artifact. Subjects were scanned as they played two conditions: regular DDR and DDR without rhythm (audio off).

Gamers performed better (gamescore) with audio than without audio. A large amount of neural activity was present in motor association areas and working memory areas of the cortex including parietal association, premotor, and prefrontal in gameplay, both with and without audio. Conjunction analysis of both conditions found the largest areas of common activation were in superior parietal gyrus, motor, prefrontal, and premotor cortex, and thalamus. Gameplay with audio revealed increased activation in temporal lobe and superior prefrontal cortex. Gameplay without audio revealed increases to brainstem, insula, premotor, and prefrontal cortex.

It is possible that shared networks active in playing the game with and without audio represent neural pathways involved in specific sensory-motor control, in which temporal and spatial accuracy of button pressing is part of the goal. These networks are diversely distributed and may not function in isolation. Rhythm information may provide players enhanced information, processed by different neural pathways, and may underlie differences in gamescores. Previous studies further support this idea and argue that networks active with rhythmic stimuli may represent top-down (e.g. feedforward) neural processing. External visual cueing has been previously shown to activate bottom-up pathways involving the insula, similar to activity we found in gameplay without audio. Finally, differences in scores and feedforward neural activity may represent an interesting mechanism of meaningful play within rhythmic games. Understanding how players improve performance in the presence of combined feedback and feedforward information should be taken into consideration for development of other game genres.