Send the link below via email or IMCopy
Present to your audienceStart remote presentation
- Invited audience members will follow you as you navigate and present
- People invited to a presentation do not need a Prezi account
- This link expires 10 minutes after you close the presentation
- A maximum of 30 users can follow your presentation
- Learn more about this feature in our knowledge base article
PSY 838 Functional Imaging of Cognitive Processesing: Execut
Transcript of PSY 838 Functional Imaging of Cognitive Processesing: Execut
Damasio et al. (1994)
Strategic Control of Memory
Interest in executive functions rooted in studying memory - earliest neuroimaging studies of executive control and PFC used WM/episodic memory tasks.
Neuroimaging of executive function: attentional control and selective attention
"What neural processes ecome engaged when inappropriate actions must be suppressed...?"
7 Distinct Categories of Executive Function
Strategic Control of Memory
Summary & Conclusions
Socially Well Adapted
Appeared to recover fully
No Movement, Speech, Learning or Memory Impairments
No Social Conventions or Responsibility
"In the words of his friends and acquaintances, “Gage was no longer Gage”."
"... the most prized of all our mental faculties..."
What functions are "executive"?
How many distinct functions are there?
Importance of frontal lobes. Development of tasks (WCST, TOL, go no-go, etc.) assessing executive function and corresponding brain areas
Importance of PFC in active maintenance functions (lesion studies and animal electrophysiology studies)
SAS theory (Norman & Shallice, 1986)
Top-down biasing effects of goal information through a control mechanism
Control mechanism located in frontal lobes, specified tasks most dependent on frontal integrity:
suppression of habitual response
novel, unpracticed tasks
dangerous, technically difficult tasks
Cognitive models by Cohen and colleagues influenced by SAS
Theoretical Models of Cognition
Baddeley's model of WM (Baddeley, 1986)
Phonological loop & Visuospatial sketchpad - "mental blackboard" for complex cognition
Central Executive using the stored information as necessary in task-related processes
Single or multiple mechanism(s)?
Contributions to executive control research are very recent...
Many WM or episodic memory tasks may actually tap into executive processes more.
Neuroimaging studies focussed on memory tasks that look at strategic control of mnemonic processes:
recency memory (which is most recent) - DLPFC (esp. RH)
source memory (specific context) & prospective memory (retrieving a goal at a specific time) - Anterior PFC
Semantic memory (e.g. TOT) - DLPFC & ACC
Control processes of left PFC involved in encoding memories - possibly updating task or instructional contexts
LH PFC responsible for elaborative phonological and semantic processing of verbal stimuli
Left inferior PFC anterior/posterior division: controlled access to phonological and semantic representations
Manipulation of information maintained in WM
DLPFC active in spatial manipulations, e.g. alphabetic reordering
Varying difficulty of different tasks
Generic manipulation processes vs. Special mechanisms
Is WM updating a unique executive process? Or part of the more generic process of inhibiting WM contents?
Inferior PFC shows 'transient' activity
n-back task, WCST when card sorting rule updated
Recent negative Sternberg task
WM updating: deactivating previously stored info to enable new info to gain access to storage
What neural control processes are used to supress interference from incongruent word-name info?
1. Incongruent vs. neutral
2. Incongruent vs. congruent
3. Frequency of trial types
4. Comparing different incongruent stims
5. Comparing different types of relevant and irrelevant stimulus dimensions or response modalities
6. Practice effects
ACC and lateral PFC are more active during conditions of high interference.
activity level of lateral PFC
The distinction between source and site:
Source of Stroop attentional control - Lateral PFC
Site of attentional modulation - Extrastriate cortex
PFC and pre-cuneus may form attention enhancement network
Parietal and extra striate cortex may form attention inhibition network
In the real world
In the lab
Disinhibition syndrome in neuropsych patients, caused by TBI
go-nogo task, Luria's tapping task, stop-signal task, anti-saccade task
Findings & Potential Problems with go-nogo
ACC and RH PFC/Parietal cortex activation associated with inhibition, i.e. nogo
Some have observed bi-lateral activation
Some have suggested response frequency of go and nogo trials to drive neural activity
Which category (go or nogo) is more salient to participant
Activation of right inferior PFC on successful trials - provding support for other findings of inferior PFC in inhibition
Support from neuropsych: damage to right inferior PFC associated with deficits in stop-signal task. Extent of damage correlated with degree of inhibitory deficit.
"free selection" "willed action"
Tasks that elicit responses that aren't heavily influenced by environment/context illustrate our "truly voluntary or willed" behaviour, which a hallmark or executive control.
Tasks where conditions allow for many equally correct responses to a stimulus - trigger an endogenous control mechanism.
Verb generation tasks
Stem completion tasks
Neuroimaging of these tasks finds activation of medial frontal regions - ACC, pre-SMA, and left lateral PFC
Nouns with many possible associations, i.e. more choices, increased activation of left inferior frontal cortex and ACC
Verbs produced that were only weakly associated with the noun and nouns that had a strongly associated verb activated ACC but not inferior frontal cortex.
When stim cues are insufficient left inferior frontal cortex may provide goal-based facilitation of semantic or lexical retrieval
ACC may detect the high degree of response conflict by the target noun (as in stimulus-response interference)
Random number generation paradigm (Baddeley et al., 1998): THE MOST EXTREME FORM OF VERBAL UNDERDETERMINED RESPONSE TASK
must verbally generate, quickly, a non-repetitive, non-stereotyped sequence of digits
DLPFC activation activity observed, suggesting DLPFC serves as a response controller (monitoring and suppressing responses from habit)
Support from TMS studies:
*no behavioural changes when RH deactivated
Pre-SMA instead of DLPFC?
Pre-SMA involved in endogenous generation of potential responses and DLPFC involved in the selection of an option: Lau et al. (2004) found DLPFC activity to be equal in a random selection task and a cued selection task but ("a region" of) pre-SMA was only engaged in random selection.
Dynamically monitoring and adjusting behaviour is important in many task situations to optimally achieve goals.
1) environmental or internal cues indicating performance success or failure
2) maintenance and integration of these cues over time, assessing and confirming subtle trends
3) performance info —> an adaptive adjustment of cognitive or response strategy
Performance monitoring and executive control classically studied with the Wisconsin Card Sort Task (WCST)
PFC damaged patients/patients with executive function deficits show a failure to appropriately use feedback to adjust strategies.
Neuroimaging studies of WCST find PFC to be involved in global performance/feedback processing
(selective) bilateral activation of inferior PFC for trials with negative feedback
feedback related activation in ACC, inferior and dorsolateral PFC
The ERN (error related negativity) was identified in ERP - occurred after an error (no feedback/internal and external feedback) in simple cognitive tasks
Poor spatial resolution of ERP suggested source of ERN around ACC - MRI confirmed
Concluded ACC detects errors and responds by correcting behaviour, e.g. slowing responses
Conflict-monitoring theory suggested that ACC’s role may be more general - detect conflict when generating a response, even when eventual response was correct (explaining why ACC activity is observed across response inhibition, stim-response interference, and undetermined response tasks)
Error detection is a special case of conflict
Importantly, ACC feeds into lateral PFC, where a change in task goals, response speed, or attentional bias is implemented
Feedback related brain activity within gamling or decision making task involving monetary rewards/penalties.
Orbitofrontal cortex (OFC) involved in trial by trial monetary feedback
OFC may be important for keeping track of the valence of such feedback
Part of executive control is the ability to internally represent task sets or task rules - goals within task context.
Best way to study task management processes?
Two paradigms to experimentally explore task management:
1. Task switching
2. Dual task coordination
Requires rapid switching among (at least) two tasks in either uncued-predictable/cued-random sequence
Performance decreases on trials where the task switches (vs. repeats)
a distinct process is required to accomplish the “switch”
Mixed evidence from event related fMRI supporting increased lateral PFC activity during task switching (vs. task repeating)
more reliable evidence for superior parietal cortex activation during switch related activity.
Greater PFC activity when:
random, rather than predictable, task sequences
following and explicit preparatory cue
suppressed task has to be activated again
task set representation complexity increases —> more complicated task rules
Anterior PFC (rather than DLPFC) showed sustained activation during task switching rather than single task
'mixing cost', the performance cost due to multitasking
a higher-order sub goal monitoring OR preparatory attentional process, i.e.detects cues indicating a switch
Parietal cortex showing most reliable effects of all switching conditions (commonalities > differences)
different perceptual features
different perceptual dimensions
Dual task performance requires the coordination, scheduling, and segregation of task representation, not just rapid updating
Regions involved in dual tasking might be the central task coordinator
No definitive demonstration of selective dual task related activity
How, Why, and under what conditions do two tasks interfere with each other?
1. Two tasks will interfere with each other when they involve the same regions of cortex
Activation in a task related area is less than the sum of activity in each of the single task conditions… “subadditive interaction”
A constraint on task related activation due to inhibitory processes, limited processing capacity, or a combo.
2. The processing stage where interference occurs
PRP (psychological refractory period effect): Decreased RT when a 2nd task is introduced while 1st task is still in a particular stage (but not if it has already passed that stage)
Increased activity in Right inferior PFC during high overlap, but not low overlap
cross task interference
Two strategies have evolved for neuroimaging research in this domain:
1. Use or adapt classic neuropsych or intelligence tests: tower of hanoi, syllogisms, etc.
2. Develop new experimental paradigms (or novel variants of current tasks) that make it easier to isolate/manipulate variables
Get at the essential cog elements of standard tasks using a simplified framework
TOL (Tower of London) planning task (most popular)
All logical problems seem to activate lateral PFC, e.g. inductive & deductive reasoning, transitive inference
dissociating judging logical validity (deductive) vs. judging plausibility (inductive)
A review (Christoff & Gabrieli, 2000) suggests a common component of these types of tasks is that they require internally generated monitoring or evaluation representations and that many of these tasks activate anterior PFC
e.g. In TOL one must internally generate several possible moves and decide which is best to reach their goal
Cognitive branching: "the process of maintaining information in WM while performing a second task, such that the two sources of info can be evaluated or integrated” (Koechlin et al., 1999)
Planning/Novel problem-solving/Abstract reasoning:
The things that set us apart from other animals!
High overlap with area involved in visual spatial WM tasks. BUT tower of london associated with increased activity of anterior PFC
widespread activation in a network of parietal, prefrontal, and medial frontal regions
activity in DLPFC and anterior PFC
* similar results in other studies with different problem solving tasks
activation of left inferior PFC during deductive
activation of left DLPFC in inductive
strong demands on verbal WM and phonological loop
evaluation of propositions against previous knowledge, engaging specific executive monitoring processes
How are the negative/affirmative semantic and syntactic aspects of sentences encoded in the human brain?
Using fMRI, evaluted language processing by having participants read negative and affirmative sentences.
Is it possible to use brain activation to identifysemantic features of different polarities?
Used multi-voxel pattern analysis (MVPA)
MVPA extract differences between temporal and spatial characteristics from recorded brain signals (Behroozi et al., 2011; Mitchell et al., 2004)
Reliably used in other domains: spatial cognitive tasks (Haxby et al., 2001)
Different levels of sentence construction processed in distinct areas:
Syntactic: inferior frontal (Caplan et al., 1998; Just & Carpenter, 1996)
Semantic vs. non semantic decisions: left prefrontal, left temporal (Wagner et al., 1998)
Orthographic and phonological: left inferior frontal cortex (Fiez & Petersen, 1998; Rumsey et al., 1997)
Left ventral occipitotemporal: Word recognition (Wandell, 2011)
Right PFC: Open ended problem solving (Vartanian & Goel, 2005)
Right ventral lateral PFC: Word selection
Right DLPFC: Syntactic sentence comprehension
Left DLPFC: Semantic sentence comprehension (Manenti et al., 2008)
Left premotor cortex: Negative sentences
Right supramarginal gyrus: Affirmative sentences (Christensen, 2009).
Machine Learning Methods
fMRI system & preprocessing
Data collection: 3.0 T GE scanner
Spiral Pulse sequence: T2* weighted data, TR=500ms, TE=18ms, 50 degree flip angle, in-plane resolution=3.125mm, slice thickness=3.2mm
Data normalized using Talariach coordinates
Used structural/anatomical ROIs, i.e. based on sulci, gyri and other visible features.
20 trials: 10 affirmative, 10 negative, presented randomly
Decoded the different sentence polarities from whole brain data through a feature selection method, ignoring voxels with no useful info
Multiple regions contain info about sentence polarity, ROI analyses indicate most info found in right and left DLPFC
Sentence polarities can be decoded in right DLPFC
Greater accuracy for negative sentences than for positive in right DLPFC
Negative sentences have more syntactic structure
Cannot explain the semantic structure of negative sentences
Negative and positive accuracy in right DLPFC
Negative and positive accuracy in left DLPFC
Accuracy of decoding of negative and positive sentences accross subjects in left and right DLPFC
Mind Wandering: The occurence of thoughts unrelated to the current task. Or, spontaneous thought unrelated to a focal task; task un-related thoughts; stimulus independent thought.
Common, can be determental to task performance
Hypothesized to have benefits, known to interfere with tasks, such as reading comprehension
Experiments using particpant subjective reports find activity in default network areas (associated with self-referential thought) and in dorsal ACC and DLPFC (associated with executive control).
These areas found to be functionally connected during mind wandering (Christoff, 2012)
A meta-analysis of neuroimaging studies of discourse processing (Ferstl et al., 2008) indicate posterior cingulate cortex, medial prefrontal cortex, temporoparietal regions (near angular gyrus) - Mind wandering and reading comprehension likely share many cognitive processes.
Self-explanation: putting the text in their own words, using prior knowledge, bridge across sentences and paragraphs
Rereading & paraphrasing: "other less effective strategies"
Neural correlates of cognitive control and discourse comprehension are active during effective strategies
Domain general network: DLPFC, ACC, pre-SMA, anterior insular cortex, PPC, inferior frontal junction, dorsal pre-motor cortex (Chein & Schneider, 2005; Duncan, 2010)
During rereading: low engagement of cognitive control network, mind wandering would lead to increases in ACC & DLPFC activity - ROIs in cognitive control.
During paraphrasing and self explanation: Cognitive control network more engaged, therefore no activation of cognitive control ROIs.
Sought to replicate prior findings from studies contrasting the same 3 learning strategies.
Participants self report mind wandering while reading - Examine brain regions associated with mind wandering.
Participats: fMRI n=15, behavioural only n=24 (increase statistical power of pretest/posttestcomparison)
3 strategies x 3 texts (3 intro physics topics - 15 paragraphs each - about 1 diagram/paragraph)
"How often did you catch yourself zoning out...?"
15 MC questions - incentive! $0.25 for each correct answer!
2 sessions separated by 2-5 days, fMRI 2nd session (1st session in mock MRI)
Data acquisition & analysis
Siemens 3T - Functional data co-registered to volume anatomical images using Siemens MP-RAGE sequence
T2* weighted EPI pulse sequence, TE=29ms, TR=2000ms, slice thickness 3.5mm, flip angle 76, in-plane resolution=3.28x3.28mm
Structural & functional images transformed to Talairach coordinates
Self explanation led to greater learning than rereading t(36)=2.30, p=0.28
Rereading did not differ from paraphrasing, paraphrasing did not differ from self explanation
More time spent rereading than paraphrasing or self explanation (43.2s/37.93/37.51s)
Mind wandering was rated lower for self explanation than for rereading (p=0.005) and paraphrasing (p=0.006) - rereading and paraphrasing did not differ
Differences in activation associated with reading strategies vs. previous study, i.e. replication
self explanation > rereading contrasts
no overlap: right superior/middle occipital gyri & fusiform gyrus
Cognitive control network in LH and right ACC & pre-SMA showed differences between rereading and self explanation and paraphrasing - no difference between self explanation and paraphrasing
Mind wandering regions correlated with participant self reports: Increased activation in control network (right IFJ, right dPMC, right & left ACC) when rereading compared to self explanation and paraphrasing
Mind wandering during rereading associated with increased activation of right-lateralized cognitive control areas
rereading task minimally engages processes that are primarily left lateralized
Much of what is thought about during mind wandering is planning and upcoming social events/interactions, requiring controlled processing
Some ambiguous findings, e.g. IFJ activation not affected by 3 strategies, but is modulated by mind wandering during rereading - task dependent?
Strong replication of prior work of strategy differences (Moss et al., 2011)
Additional activation of fusiform gyrus most likely because of inclusion of pictures in current study
There is a diverse set of tasks/studies assessing executive functions
As executive control probably involves widespread interaction and coordination of processing from multiple systems, complex analysis techniques are needed (SEM, partial least squares, dynamic causal modelling, etc.)
Lateral PFC (particularly DLPFC) & ACC are reliably engaged during almost every task requiring executive control
Is this the "site" of the central executive?
The future: figuring out the division of labour: What is happening in PFC that makes it critical to executive control? Why PFC?