Guest Lecture Notes

Sunday, December 4, 2022 11:59 PM

Dual Mechanisms of Control: A refined cognitive control framework that examines why we see variability within
people (wrt cognitive control)

Cognitive Control: combines several factors such as WM, mental brakes (inhibiting behaviours that deter you from
achieving a goal), and cognitive flexibility etc.

Despite how important cognitive control is, and how it allows us to regulate and coordinate thoughts and actions,
there is a core paradox within psychological and neuroscience literature: Why is cognitive control important, yet so
vulnerable to internal and external factors, and why does it differ between people and populations?

Cognitive control may not be just be one construct, it may be differentiated into two different modes. This is where
the DMC comes in.

Dual Mechanisms of Control

• Proposes two types of control: reactive and proactive

• Proactive Control: refers to when we are actively aware that something will interfere with our ability to stay on
task. It depends on contextual information so that we can anticipate and prevent any interference before it Note: IPS is not
occurs. It also serves to guide information processing system (IPS) towards goal relevant information before a formal short
any stimulus that may interrupt the IPS. form
• Reactive Control: Attention is recruited as a late correction mechanism and only mobilised as needed.
Therefore, it is only adjusted in real time after interference has come up or as it is detected. Reactive control
is like a "defense mechanism" that only occurs after goal achievement has been interrupted.

• There are complementary trade-offs between reactive and proactive control: using a proactive control too
early may end up in you deploying cognitive resources that need to be used elsewhere, therefore you might
end up having to use a reactive approach due to the several stimuli you need to focus on

Neural Correlates

Proactive Reactive
Sustained Lateral PFC activation Transient activation
activation
Active maintenance Ability to stay on task Detection of
of task goals interference (by Note: areas in yellow are
anterior cingulate expected to be active during
cortex) corresponding control
strategies
Phasic Brain Area/Signal Associative and
Dopaminergic- Episodic associations
mediated gating in posterior cortical or
signal MTL region

Variability
1. Intra individual: situational differences that result in either one of proactive and reactive strategies by the
same person
2. Inter individual: stable individual differences that lead to people being biased differently to either type of
strategy
3. Between groups: differences between population

Intra Individual
• An important component of cognitive control is working memory
• WM is responsible for managing and manipulating information
• WM capacity is limited by interference. Therefore if there is more interference, WM is hindered more
• The ability to control interference is also a strong predictor of WM capacity
• In situations where interference can be predicted, you are more likely to engage in proactive control
• In situations where interference is unexpected, you are more likely to engage in reactive control RAPM is a non-
verbal abstract
• Burgess and Braver (2010) reasoning test.
• Participants: Used as a
○ N=22 measure of fluid
○ Upper and Lower Quartiles based on their performance on Raven's Advanced Progressive Matrices intelligence.
(prev. study)

• Method:
○ Under fMRI
○ Recent probes task
○ Participants first see a memory set that contains several items to remember
○ Followed by a brief delay
○ They then see a recent probe item
○ They are asked to identify if the recent probe item was present in the memory set (positive probe) or
absent (negative probe)
○ Researchers manipulated interference expectancy
• In the low expectancy condition, only 20% of the recent probe items were negative
• In the high expectancy condition, 80% of the recent probe items were negative

• Results:
○ Those with higher fluid intelligence made fewer errors compared to those with lower fluid intelligence
 ○ Low fluid intelligence participants showed higher interference effect compared to the higher fluid
intelligence Interference effect = slower reaction
○ Stronger interference effect seen for recent negative probes compared to novel negative probes

○ In the image above, it is shown that the left lateral prefrontal cortex was more engaged for recent
negative probes compared to novel negative probes
○ This was also seen following the presentation of the probe, which suggest a reactive control
mechanism (due to transient PFC activity that is expected during reactive control (see Neural
Correlates table above))
○ Results also showed that the low expectancy condition was primiarily associated with interference
effect at the time of the probe, however within the high expectancy condition, we see probe-related
activation which indicated anticipatory and global proactive control effect

Inter-Individual Variation: Anxiety

• DMC assumes that there may be stable individual differences that lead to bias towards either proactive or
reactive control, and what the preferred strategy might be in performing tasks with difference cognitive
control demands
• The utilisation of proactive control would relate to cost-benefit trade offs that relate to the efficacy and
efficiency of maintaining goal-related information
• People who are more anxious may be more prone to reactive control in many situations as a way to adapt
and control the environment around them.
• Reactive control may be more desirable in people with anxiety as they are able to direct their attention to
unpredictable threats as and when possible, and actively monitor the situation they are in

• Fales et al., (2008); Gray, Chabris and Braver (2003)

• Method:
○ High and low-anxious participants completed an n-back task while under an fMRI
○ 3 back match as shown:

○ Proactive control is measured through how working memory mechanisms are recruited, i.e. how task
stimuli is maintained during the trial and how efficiently a target response is provided
○ Three types of trials:
• Target: is a 3-back match
• Lure: non-target, but they have been presented at least once previously (≠ 3 back match)
• Nonlure: non-target, not previously presented

• Results:
○ High anxious groups showed significantly reduced sustained activity in cognitive control regions but
increased transient activity during the task trials
Fales et al (2008)

○ High anxious: Yellow in sustained, Blue in target/lure/nonlure (transient)

○ Right ventrolateral PFC has been linked to response inhibition and interference, and greater activation
during non-target may be sensitive to evaluating conflict
○ Therefore, results suggest that low anxious people constantly engage right ventrolateral PFC, but high
anxious people recruit more actively during trials in a transient manner (therefore reactive)

Inter-Individual Variability: Reward Sensitivity

• Dopamine is linked to reward sensitivity
• When motivation is high or there are rewards involved, there is increased value of goal achievement
• DMC predicts that proactive strategy would be dominant under context of rewards because proactive control
would allow you to maintain and update task goals ahead of time, and globally facilitate performance in a
 reward-motivated context

• Jimura, Locke, and Braver (2010)

• Method:
○ fMRI study
○ Reward block and non-reward block
○ Participants performed a working memory task in which a reward cue was presented at the beginning
and following the cue they were presented with a word list
○ There was then a delay
○ Then they were shown a probe and asked if the probe was present in the word list
○ Participants were asked to be both accurate and fast in their responses
• Results:
○ Reaction times were faster in the reward context than in the non-reward context by almost 20%
○ Suggests that cognitive performance is modulated by potential rewards (consistent with previous
findings about motivation and WM)
○ Highly reward-sensitive has greater performance enhancement specifically due to the motivational
context and not just the reward they were getting
○ Shift towards proactive control in the reward context — seen through increase in sustained lateral PFC
activation, decrease in transient lateral PFC activation
○ Transient Activation: the effect of context was significant in the late trial with robust activity in the
reward negative context and decrease in the reward positive context (see graph below)

○ Reward sensitivity was positively correlated with the increase in sustained activation in the reward
context (and additional transient activity during the EARLY trial period)
○ This is characteristic of taking a proactive control (changing from reactive) based on context

Between-Group Variations: MDD

• MDD symptoms include: persistent low mood, anhedonia
• Difficulties with inhibiting attention to task-irrelevant negative information
• Brain regions involved: DLPFC, dorsal ACC, inferior parietal cortex, and striatum
○ Studies have found increased and decreased activity in these regions within MDD patients
○ So while these areas are involved, we are not sure how they are involved
○ Inconsistent results may be due to looking at cognitive control as one model instead of using the DMC
and looking at proactive and reactive controls individually

• Vanderhasselt et al., (2014)

• Aim: investigate whether deficits in inhibiting negative information in patients with MDD were related with
proactive and reactive controls

• Hypothesis: MDD patients would have increased response latencies when encountering sad faces, which
would reflect selective impairment they have inhibiting responses to negative information

• Method:
○ EEG
○ Task: Cued emotional conflict

○ Each trial starts with one of the cue words

• Actual: response with the actual emotion of the upcoming target face
 • Opposite: response with opposite emotion of upcoming target face
• Press: press a third button unrelated to emotion of target face

• Results:

○ MDD patients were selectively slower when having to actively inhibit a sad face and having to use the
opposite response key
○ Consistent with previous studies that show that MDD patients have a difficulty with overwriting
responses when it comes to negative stimuli

○ Anterior Cingulate Cortex (performance monitoring and conflict) was more active in MDD patients for
opposite-sad condition — suggests that they have a difficulty in resolving conflict
○ Specifically:
• Study looked at Contingent Negative Variation Amplitude of the EEG — CNV amplitude is
significant as abnormal activity is observed in patients with MDD
• This reflects difficulty during effortful processing
• Researchers found that MDD patients did not show difference in CNV amplitude compared to
controls, which suggests they have difficulty engaging in proactive control

Between-Group Variations: Schizophrenia

• Cognitive control and dysfunction in lateral PFC
○ Specifically dysregulated input from midbrain dopamine system
• Patients with schizophrenia have dysregulated DA-PFC interactions
• Patients with schizophrenia may show a shift from proactive to reactive control
○ Dopamine is important for proactive control (see Neural Correlates table above)
• Edwards, Barch, and Braver (2009)
• Aim: investigate proactive and reactive controls in patients with schizophrenia

• Participants: 22 patients with schizophrenia and 14 healthy controls

• Method:
○ fMRI
○ AX-CPT Task — AX-Continuous Performance Task)
• 70% AX (target trials)
• 30% AY, BX, BY (non-target trials)

• First letter appears, and that information is maintained

• Then second letter appears, and participant is asked to response whether the combination is the
target pair.
• Here, any combination that is not AX is non-trial

○ According to DMC, there would be sustained activation in the lateral PFC to stay on task and as a sign
of proactive control
○ Completed AX-CPT twice:
• Pre and post training

• Results:
○ Pre-training schizophrenia patients had a greater number of errors and slower response time for both
AY and BX trial types
○ Researchers calculated proactive behavioural index (standarised number about whether someone is
engaging in proactive control)
○ Based on calculation, pre training healthy patients showed higher proactive control compared to
schizophrenia in terms of reaction time
 ○ Post training, schizophrenia patients saw improvement in proactive control wrt reaction time AND error
rate

○ Within the lateral PFC, there was a pattern of proactive cognitive control post-training
○ At baseline, patients with schizophrenia showed pattern that was consistent with reactive control
○ After training, it normalised their cognitive control patterns and shifted them back to proactive control
similar to control participants (see table below)

○ There was also correlation seen wrt clinical symptoms

○ Schizophrenia patients with disorganised symptoms had a greater benefit from training
○ This suggests that there was an association between clinical symptoms and training effects wrt
engagement in proactive control following training

Between-Group Variation: Aging

• Cognitive control and dysfunction in lateral PFC
○ Dysregulated input from midbrain dopamine system also seen in aging individuals
• Paxton et al., (2008)
• Aim: Investigated older v younger adults

• Method: Participants completed a variant of the AX-CPT

• Results:
○ Older adults showed increased reaction times and errors on the BX trials than younger adults
○ Older adults showed decreased reaction times and errors on the AY trials than younger adults
○ These differences in reaction times were greater in BX trials than AY trials
○ This shows older adults show greater deficits with goal maintenance compared to younger adults
○ Reduction in proactive control in older adults (shown through decreased cue related activity)
○ Also an increase in reactive control in older adults (shown through increase probe related activity)
(see graph below)

○ Although old adults and schizophrenia patients are pretty different, they have similarities with impaired
cognitive control performances
○ Results suggest, from DMC perspective, that similar patterns of proactive and reactive control can be
found in diverse groups that show different cognitive control impairments
 ○ A shift from reactive to proactive (and vice versa) may serve as a common pathway that might explain
diverse set of pathologies that you might see in different groups

Conclusion
• DMC framework provides a new way to look at cognitive controls focusing on intrinsic variability within
individuals, between individuals, and between groups
• It is not limited to cognitive traits like WM or fluid intelligence
• It considers non-cognitive traits like reward sensitivity and anxiety