Intended actions and unexpected outcomes: automatic and controlled processing in a rapid motor task

Abstract

Human action involves a combination of controlled and automatic behavior. These processes may interact in tasks requiring rapid response selection or inhibition, where temporal constraints preclude timely intervention by conscious, controlled processes over automatized prepotent responses. Such contexts tend to produce frequent errors, but also rapidly executed correct responses, both of which may sometimes be perceived as surprising, unintended, or "automatic". In order to identify neural processes underlying these two aspects of cognitive control, we measured neuromagnetic brain activity in 12 right-handed subjects during manual responses to rapidly presented digits, with an infrequent target digit that required switching response hand (bimanual task) or response finger (unimanual task). Automaticity of responding was evidenced by response speeding (shorter response times) prior to both failed and fast correct switches. Consistent with this automaticity interpretation of fast correct switches, we observed bilateral motor preparation, as indexed by suppression of beta band (15-30 Hz) oscillations in motor cortex, prior to processing of the switch cue in the bimanual task. In contrast, right frontal theta activity (4-8 Hz) accompanying correct switch responses began after cue onset, suggesting that it reflected controlled inhibition of the default response. Further, this activity was reduced on fast correct switch trials suggesting a more automatic mode of inhibitory control. We also observed post-movement (event-related negativity) ERN-like responses and theta band increases in medial and anterior frontal regions that were significantly larger on error trials, and may reflect a combination of error and delayed inhibitory signals. We conclude that both automatic and controlled processes are engaged in parallel during rapid motor tasks, and that the relative strength and timing of these processes may underlie both optimal task performance and subjective experiences of automaticity or control.

Keywords: ERN; MEG; automaticity; beta oscillations; frontal theta; motor cortex; response inhibition; response switching.

Figure 1

(A) Response switching task. Digits were presented every 1150 ms for a duration of 250 ms, followed by a stimulus mask (“&”) for 900 ms. Overall target digit (“3”) probability was 20%. (B) Response Time (RT) means and standard error for all trial types for the within-hand (WH) and between hand right-to-left (BH) tasks.

Figure 2

Mean RT for fast, slow correct and incorrect responses to target digits and the immediately preceding (Pre) and following (Post) defaults (non-target) responses for WH and BH tasks. Slow and fast switch trials correspond to the fastest 1/3 and slowest 1/3 RTs, respectively. Error bars indicate mean standard error, and gray bars the mean ±1.0 standard deviation RT for all default trials.

Figure 3

(A) Group images of event-related sources in motor cortex (precentral gyrus). Left and middle images show activity for default (right index) responses for the within hand (WH) and between hand (BH). Image at far right shows motor cortex activity for switch trials (left index finger). (B) Averaged source waveforms of brain activity in the left and right motor cortex for all conditions. For switch trials (green traces) significantly larger amplitude was observed prior to movement onset (^*p < 0.02). Polarity reversals ipsilateral to the switch hand for successful switch trials are indicated by red arrows.

Figure 4

Theta band (4–8 Hz) activity preceding correct switch responses. (A) Significant activations in the right middle frontal gyrus for all tasks during the pre-movement period (−0.5 to 0 s) relative to baseline (−1 to −0.5 s). (B) Time-frequency plots of induced source activity (1–50 Hz) for the peak locations shown in (A) for the within hand (WH) and between hand (right to left) conditions. (C) Time course of total power within the theta (4–8 Hz) band comparing switch and default trials (left plot). Plot on right shows time courses for subsets of “fast” and “slow” switch trials for both WH condition (solid traces) and BH conditions (dotted traces). Shaded bars indicate the time range of stimulus onset (mean ± standard error) for the WH condition.

Figure 5

Time course of total power in the beta (15–30 Hz) band in motor cortex in both bimanual tasks. Contralateral motor cortex (upper row) and ipsilateral MI (lower row) correspond to motor cortex contralateral and ipsilateral to the default response hand, respectively. Correct trials are shown in green and error trials are shown in red. Shaded rectangles show the time range (±1.0 SD around the mean) of visual cue onset for the correct (green shading) and error (red shading) trials. Dotted lines show the difference (error–correct) in beta power. (A) Right to left switch condition (contralateral MI = left motor cortex). (B) Left to right switch condition (contralateral = right motor cortex). Solid grey lines indicate period during which differences between correct and error trials was significant (paired t-test, p < 0.05).

Figure 6

(A) Event-related beamformer source images showing error-related brain activation a latency of approximately 90 ms following the incorrect button press in the region of the anterior cingulate cortex for between hand (right-to-left) and within hand switch tasks. (B) Group averaged time course of source activity relative to movement onset (t = 0 s) for peak activations shown. Peak activity for error trials was significantly different from default trials for both tasks (^*p < 0.01, ^**p< 0.001, paired t-tests, corrected).

Figure 7

Error related theta band activity. (A) Source images of theta (2–8 Hz) power (p < 0.05) for the Error > Default contrast for within hand (WH) and between hand (BH) tasks. (B) Time-frequency plots for virtual sensors corresponding to the peak locations. (C) Time course of total power in the theta band for error trials (solid lines) and default trials (dotted lines) for both WH and BH conditions. Shaded rectangles indicate the approximate time range (±1.0 standard deviation from the mean) of visual cue onset. Vertical bars indicate standard error across subjects. Difference in peak theta power at latency of 150 ms following button press (indicated by asterisks) between error and default trials were highly significant for both WH and BH conditions (p < 0.0004 and p < 0.00003, respectively, paired t-tests, corrected).