Network Dynamics Underlying Speed-Accuracy Trade-Offs in Response to Errors

Yigal Agam,Mark Vangel,Dara S. Manoach,Kara A. Dyckman,Caitlin Carey,Jason J. S. Barton,Adrian K. C. Lee,Gilles Pourtois

doi:10.1371/journal.pone.0073692

Abstract

The ability to dynamically and rapidly adjust task performance based on its outcome is fundamental to adaptive, flexible behavior. Over trials of a task, responses speed up until an error is committed and after the error responses slow down. These dynamic adjustments serve to optimize performance and are well-described by the speed-accuracy trade-off (SATO) function. We hypothesized that SATOs based on outcomes reflect reciprocal changes in the allocation of attention between the internal milieu and the task-at-hand, as indexed by reciprocal changes in activity between the default and dorsal attention brain networks. We tested this hypothesis using functional MRI to examine the pattern of network activation over a series of trials surrounding and including an error. We further hypothesized that these reciprocal changes in network activity are coordinated by the posterior cingulate cortex (PCC) and would rely on the structural integrity of its white matter connections. Using diffusion tensor imaging, we examined whether fractional anisotropy of the posterior cingulum bundle correlated with the magnitude of reciprocal changes in network activation around errors. As expected, reaction time (RT) in trials surrounding errors was consistent with predictions from the SATO function. Activation in the default network was: (i) inversely correlated with RT, (ii) greater on trials before than after an error and (iii) maximal at the error. In contrast, activation in the right intraparietal sulcus of the dorsal attention network was (i) positively correlated with RT and showed the opposite pattern: (ii) less activation before than after an error and (iii) the least activation on the error. Greater integrity of the posterior cingulum bundle was associated with greater reciprocity in network activation around errors. These findings suggest that dynamic changes in attention to the internal versus external milieu in response to errors underlie SATOs in RT and are mediated by the PCC.

Highlights

Trial-by-trial variability in reaction time (RT) is ubiquitous in human behavior, yet its neural basis is poorly understood
Simple integer weightings, which represented our idealized predictions for RT across trial positions based on the speed-accuracy trade-off (SATO) function (21 for 2PRE, for 1PRE, for ERR, 2 for 1POST and 1 for 2POST), correlated strongly with RT (t(34) = 19.0, p,10210) indicating that RT across trial position was consistent with the SATO function
We examined whether activation during antisaccades across trial positions in the default and dorsal attention networks correlates with RT and the idealized SATO predictions

Summary

Introduction

Trial-by-trial variability in reaction time (RT) is ubiquitous in human behavior, yet its neural basis is poorly understood. SATOs are generally studied by manipulating participant instructions to emphasize either the speed or accuracy of responding. These manipulations are thought to lead to strategic changes in decisions about how much evidence is required to initiate a response. Responses speed up until an error is committed i.e., pre-error speeding; [3,4] and following an error, RT slows [i.e., post-error slowing; 5], and the probability of an error decreases This pattern can be interpreted as a progression to riskier positions on the SATO function culminating in an error, which is followed by a shift back to safer, more careful responding with a greater likelihood of a correct response

Methods

Results

Conclusion