Introduction
While humans are often remarkably efficient at individual tasks like reaching or looking, a study published in PLOS Computational Biology reveals a significant breakdown in performance when these tasks must be complexly coordinated to maximize a monetary reward. In an eye-hand coordination task designed to reward speed, human observers failed to adopt mathematically optimal strategies, achieving only about 78% of the maximum possible gain. This suggests that while we are "near-optimal" in simple movements, a cognitive "cost of coordination" prevents us from perfectly integrating independent visual search and manual reaching plans.
The Coordination Problem
In daily life, our eyes and hands are "tightly coupled"—we usually look at what we are about to touch. This "hand-follows-eye" strategy is effective for routine tasks, but the researchers investigated whether humans can decouple these systems when a reward structure explicitly penalizes the time taken to find and reach a target. Using statistical decision theory, they defined an "ideal observer" strategy and compared it to human performance to see if we can truly optimize complex, interrelated motor and perceptual plans.
Methods: The Slow-Motion Task
The researchers designed a "very slow search and reach" task to amplify the temporal costs of movement and perception. Participants had to find a target among five distractors on a touch screen, with a monetary reward that decayed linearly with every second spent.
To make strategy critical, the task was made artificially slow:
- Discrimination: Identifying the target required fixating on it for approximately 1.5 seconds.
- Movement: Hand movement was capped at a slow 4 cm/s.
- Spatial Layout: Objects were arranged in two clusters: a "small cluster" (2 objects) and a "large cluster" (4 objects).
Optimal Strategy vs. Human Performance
The authors used statistical decision theory to derive the optimal strategy. Counter-intuitively, the eye should search the smaller cluster first. This prioritizes information gain over the raw probability of finding the target. After only two fixations in the small cluster, the observer knows with 100% certainty whether the target is there or in the other cluster, reducing spatial uncertainty fastest.
Furthermore, the optimal reach strategy dictates that the hand should start moving toward the "centroid" of all possible targets before the eye finds the target, updating its trajectory as the eye rules out distractors.
Key Results
The study found a clear failure of optimality:
- Efficiency Gap: The median observer achieved only 78% of the expected gain.
- Search Sub-optimality: Only 2 out of 8 participants consistently chose the optimal "small-cluster first" search strategy.
- Movement Lag: While participants initiated hand movements before finding the target, they moved significantly slower than their maximum allowed speed, losing valuable reward points.
- Dynamic Updating: Most participants did update their hand's direction as the eye ruled out distractors, showing that the systems are not entirely "locked," even if the integration is imperfect.
Conclusion: The Cost of Coordination
The authors hypothesize that the brain struggles to plan two independent trajectories (eye and hand) simultaneously. This "coordination cost" might lead participants to adopt simpler, sub-optimal heuristics to reduce cognitive load. Participants moved their hands slowly, suggesting a "wait-and-see" approach to avoid the motor cost of making sharp, high-speed corrections once the target is located. Future research may focus on quantifying this "CPU tax" the brain pays when attempting to run independent eye and hand programs.
Reference: Zhang H, Morvan C, Etezad-Heydari L-A, Maloney LT (2012) Very Slow Search and Reach: Failure to Maximize Expected Gain in an Eye-Hand Coordination Task. PLoS Comput Biol 8(10): e1002718.
Summary updated