associative learning and cognitive control eddy j. davelaar & geoff bird
TRANSCRIPT
Associative learning and cognitive control
Eddy J. Davelaar & Geoff Bird
Incongruent trials (><>) slower than congruent trials (<<<) flanker effectFlanker effect smaller if the previous trial was an incongruent trial Gratton effectExplained by a computational model that assumes that response conflict is monitored by the ACC, is increased for incongruent trials, and sharpens attention on the next trial.
Botvinick, et al., 2001Jones, et al., 2002Cho, et al., 2002Botvinick, et al., 2004Yeung, et al., 2004
However, the Gratton effect (the interaction) is only observed when target/response repeats across trialsPriming account of the Gratton effectMayr, et al., 2003; Nieuwenhuis, et al., 2006
>>> >>>
>>>>>><<< <<<
<><
<><
<><<><
><>><>
1) IF learning from array to response plan THEN stronger functional connectivity for iI and cC trials in repeats compared to non-repeats2) IF conflict-modulated attention THEN stronger connectivity between conflict area and attention area
This is a DCM study: 2-step process1) localise relevant voxels2) Apply DCM to those voxels
Stimuli:C: >>>>>, <<<<<I: <<><<, >><>> N: -->--, --<--
Localisers:Decision: I minus C (DLPFC)Identity: full-array adaptationConflict: I minus C (ACC/pre-SMA)Attention: I+C-N
Design:
# nonrepeat repeat
I C I C
i 36 36 36 36
c 36 36 36 36
n 36 36 36 36
72 null events
Scanning Parameters:
2 runs 19mins each separated by structural scan. Scanner time ~60mins per subject including set-up. 14 subjects (14hr scanner time in total).
Event-related design, Stimulus + response period (1.5s) blank ISI (mean 3s, 2-4s range). Random sequence of events.
Whole cortex coverage (no cerebellum). TR 2.7s, 30slices, 3.5mm, 10% gap.
>><>>
1.5sMean 3s
Conflict area (ACC/pre-SMA)
Decision area (DLPFC)
Attention area (PPC/FEF)
Stimulus area (temporal)
WL
WC
Wi = b1(previous trial-type) + b2(target/response repetition)
Priming account: b10, b2>0Conflict account: b1>0, b2=0