A quantitative interaction between signal detection in attention and reward/aversion behavior

This study examines how processes such as reward/aversion and attention, which are often studied as independent processes, in fact interact at a systems level. We operationalize attention with a continuous performance task and variables from signal detection theory, and reward/aversion with a keypress task using variables from relative preference theory. We find that while the relationship between reward/aversion and attention is functionally invariant, a power law formulation akin to the Cobb-Douglas production function in economics provides the best model fit and theoretical explanation for the interaction. These results indicate that a decreasing signal-to-noise with signal detection results in higher loss aversion. Furthermore, the estimated exponents for the multiplicative power law suggest capacity constraints to processing for attention and reward/aversion. These results demonstrate a systemic interaction of attention and reward/aversion across subjects, with a quantitative schema raising the hypothesis that mechanistic inference may be possible at the level of behavior alone.