Conflict and Error

conflict monitoring model

built on the framework of “parallel distributed processing”

• simulated neurons and interconnected neurons into computational units that had inputs and outputs

• basically neural networks and the base for A.I

how to train a neural network to perform the Stroop task

stimuli (color or word) → response (red or green) → conflict monitoring in ACC (if the semantics and visual information don’t match up) → Control in DLPFC (encodes the rule (which one should be selected) → valid response

flanker task

similar to the Stroop task but with congruent and incongruent task

• stare at the fixation point that has arrows flanking it

conflict monitoring unit

detect the need to enhance control and resolve conflic

control unit (task representation)

implements the control through top-down biasing

set up of fMRI to map conflict and control areas

separated the Stroop task into the instruction and response period

• instruction: color or word

• response: answer

the response period during the incongruent trials should engage the conflict monitoring unit

results of fMRI to map conflict and control areas

ACC: conflict

L. DLPFC: control

AxX-CPT Task

a task that increases error and response conflict

• cue: a or B

• delay period

• probe: x or y

• response

created a response mapping that is more automatic (A→x, and B→y)

other less frequent response mapping pairs will create more response competition (conflict and error)

results of AX-CPT task

the ACC activity increased for both error and response competition trials

ERN

error-related negativity

• EEGs show a more negative signal when there’s an error

• how negative the error is correlated to the reaction time (post-error slowing so you can be more attentive)

source of ERN

likely the medial PFC

what happens to error?

there might be an “error” module to detect errors, and then implement control. similar to conflict monitoring

monitoring and control of action by the frontal lobes

saccade version of the stop signal task

• FEF: direct control of initiating or canceling a saccade

• SEF: monitored actions, error neurons

monitoring system

• ACC could act as a “monitor”: detecting situations when cognitive control is needed. suck as after an error, conflict, or competition between responses

• ACC then actives task representation in the lateral PFC, which will then engage top-down control (through hierarchical organization)

EEG synchrony to map the interaction between ACC and lateral PFC

flanker task but with letters

goals:

• characterize the pattern of neural synchrony between medial PFC and lateral PFC

• could be the mechanism that allows the monitoring unit to communicate with the attention control unit

EEG synchrony results

• error trials showed increased medial PFC activity

  • could reflect the engagement of the medial PFC monitoring unit

• increase neural synchrony between medial PFC and lateral PFC for error trials

  • theta range

  • suggests that the monitoring unit is communicating with the attention control unit through theta synchrony