1
Q
What are the basic principles of functional brain imaging?
A
Based on principles of haemodynamics (movement of blood) in the brain
2
Q
What is the basic idea of functional brain imaging?
A
- bits of the brain that are working hard (i.e. with lots of neurons firing) are metabolising faster and require more oxygen.
- Oxygen is delivered to these tissues in the bloodstream
- So, if we can measure the amount of oxygenated blood, we also have an indirect measure of neural activity in different brain regions
- More neurons firing means more blood and oxygen is needed
- Don’t need to stick electrodes in the brain (non-invasive)
3
Q
How does PET (positron emission tomography) work?
A
- Radioactive isotopes (e.g. O15) are injected into the bloodstream
- These travel to the brain and the emitted radioactivity can be measured by detectors in the scanner
- The radioactivity decays quickly (good!), but this therefore limits the length of the task to around 30 seconds
4
Q
When was PET introduced?
A
First used to study human brain function in the 1970s
5
Q
How can PET scans be used with Parkinsons?
A
Can be used to study the role of neurotransmitters such as dopamine, by using radioactive isotopes that bind to dopamine receptors – note the loss of receptors in the caudate nucleus of a patient with Parkinson’s disease
6
Q
How do fMRI (functional Magnetic Resonance Imaging) work?
A
- Scanner is a giant magnet (thousands of times stronger than the Earth’s magnetic field)
- fMRI is based on measuring the differences in the magnetic properties of oxygenated and deoxygenated blood.
- Called the BOLD response (Blood Oxygenation Level Dependent)
- No need for injections
7
Q
When were fMRI (functional Magnetic Resonance Imaging) introduced?
A
Emerged in the 1990s
8
Q
How do Anatomical Brain Image’s work?
A
- Magnetic properties in hydrogen molecules in water
- Not measuring magnetic properties in blood
9
Q
What is spatial resolution like for PET and fMRI?
A
High spatial resolution – down to mm in fMRI (NB there are hundreds of thousands of neurons in every few cubic mm!)
10
Q
What does the sluggish blood flow mean for PET and fMRI scans?
A
But sluggish blood flow response means inferior temporal resolution – mins for PET, seconds for fMRI – when neurons are firing it takes time for the blood to travel (2-3 seconds) to the destination
11
Q
What is the assumption of PET and fMRI scans?
A
Assumption: high neuronal activity -> high metabolic demand -> more oxygenated blood
12
Q
What are brain images and what do they show?
A
- Statistical map
- Images show the difference in activation (PINK) between two different brain states … in this case, Speaking > Resting
- Coloured areas are called T-values
- Can lead to a stroke if blood flow doesn’t reach all areas of the brain
13
Q
What is Subtraction and ‘Pure Insertion’?
A
- The idea that new cognitive component (A) can be purely inserted into a task without affecting the expression of the previous components (e.g., B)
- A - Activation task involving process of interest (subtract)
- B - Baseline/control task identical to A except for process of interest
- = cognitive process of interest
14
Q
Pure insertion – brain imaging example:
A
- A – watching dots move (subtract)
- B – watching stationary dots
- = regions involved in motion perception
15
Q
What did Raichle (1998) do in increasing task complexity in a neuroimaging task?
A
1: Opening eyes
2: Looking at nouns
3: Reading nouns aloud
4: Saying a verb to go with each noun (e.g. car -> “drive”)
- All the changes shown represent increases in signal (can do the opposite and decrease, swap subtraction)
16
Q
The problem with pure insertion - especially subtraction?
A
- You might not be subtracting out everything!
- A – watching a movie
- B – resting with eyes closed
- = Region(s) involved in … motion, auditory…lots of areas!
- Contrasted comparison condition must be meaningful with respect to the research question!
17
Q
What does pure insertion assume?
A
- Each cognitive component is functionally separate, and different cognitive components do not interact
- Each cognitive component evokes an ‘additional’ physiological activation compared to the contrasted condition (irrespective of the cognitive or physiological context)
- Assumes a modular brain -> limited view of brain function!
18
Q
So what do we do with our BOLD dependent variable?
A
- The same as in experimental psychology
- Deductive reasoning
- Inductive reasoning
19
Q
What is deductive reasoning?
A
-Deductive reasoning = theory – hypothesis – observation – confirmation
20
Q
What is inductive reasoning?
A
-Inductive reasoning = observation – pattern – hypothesis – theory
21
Q
Two types of inference from imaging data - what is Theory about function?
A
- Theory about function (deductive reasoning)
- “function-to-structure deduction”
- context-specific
- theory-specific
- a single experiment
22
Q
Two types of inference from imaging data - what is Observation about structure
A
- Observation about structure – pattern – hypothesis – theory (inductive reasoning)
- “structure-to-function induction” (inductive reasoning)
- context-independent
- stronger assumptions
- several experiments
23
Q
What is Function-to-structure deduction (deductive reasoning)?
A
If two experimental conditions produce qualitatively different patterns of activity over the brain, then these conditions differ in at least one function.
fMRI data can tell us about functional specialisation of the brain, and can be used to distinguish between competing cognitive theories
24
Q
Example of Function-to-structure deduction (deductive reasoning)?
A
- Distinguishing between two cognitive theories
- Recognition memory: Recollection versus Familiarity (Henson et al., 1999)
- Exposure word list: Issue Market Pencil Cook Mobile Hair Glove
- Test word list: Hair Dog Market Train Glove Rose Mobile
- At test: Is the item only familiar to you (“Know”), or can you recollect seeing it previously (“Remember”)?
25
What was Theory 1 of fMRI of Remember/Know experiment (Henson et al., 1999)?
- Theory 1: “single-process” model Hypothesis: Participants’ answers depend on “memory strength”
- > Quantitative difference between ‘Remember’ and ‘Know’
- Results: remember > know
26
What was Theory 2 of fMRI of Remember/Know experiment (Henson et al., 1999)?
- Theory 2: “dual-process” model Hypothesis: Remember judgements entail Recollection + Familiarity, but Know judgements entail Familiarity only -> Qualitative difference between ‘Remember’ and ‘Know’
- Know > remember
- Data support Theory 2: so, fMRI data can be used to distinguish between competing theories!
27
What do localisation studies show?
localisation studies tell us that the brain, and hence the mind, contain specialised mechanisms for a particular mental process.
-(Mather et al., 2013)
28
Mapping cognitive functions to brain areas
| -vision ...
-vision – visual cortex
29
Mapping cognitive functions to brain areas
| -audition ...
-audition – auditory cortex
30
Mapping cognitive functions to brain areas
| -motion ...
-motion – V5/MT
31
Mapping cognitive functions to brain areas
| -colour ...
-colour – V4
32
Mapping cognitive functions to brain areas
| -faces ...
-faces – fusiform face area
33
Mapping cognitive functions to brain areas
| -places ...
-places – parahippocampal place area
34
Different parts of visual cortex represent space in ...
... a continuous and systematic way – a retinotopic map (Sereno et al., 1995)
35
Different parts of auditory cortex represent tones of different ...
... frequencies – a tonotopic map (Formisano et al., 2003)
36
What do sophisticated localisation studies tell us?
- Sophisticated localisation studies tell us about - how the brain parses sensory input - how the brain is organised and functionally specialised
- This is very important but tells us less about how high-level cognition is performed
- However, some function-to-structure deduction can also distinguish between different cognitive theories, as we have seen in the Remember/Know experiment (Henson et al., 1999)
37
What is Structure-to-function induction?
- Inferring the engagement of a particular cognitive function based on activation in particular brain regions.
- If the same brain region is activated in two (or more) different experiments, this approach assumes that the same function is engaged in both cases
- Precise location matters: the same brain region in all relevant experiments
- Strong assumption of “systematicity” in the structure-to-function mapping: the same regions are involved in the same functions in all contexts (i.e. experiments).
38
What is the example of the Verbal short-term memory (Henson, Burgess, & Frith, 2000) testing Structure-to-function induction?
- Two probe tasks to investigate serial order:
- Item Probe: Subjects shown temporal sequence of six letters. One probe letter. Was probe in original sequence? Need to maintain Serial Order – No
- List probe: Subjects shown temporal sequence of six letters. Shown six items simultaneously. Was the original sequence in same order? Need to maintain Serial Order – Yes
39
What were the results of the Verbal short-term memory (Henson, Burgess, & Frith, 2000) testing Structure-to-function induction?
- List task > item task
- Dorsal left premotor cortex: Greater fMRI activity for List probe task
- This region had been implicated in timing in other imaging tasks including finger tapping (Catalan et al., 1998)
- Interpretation? Timing is involved performing serial order tasks -> reverse inference
40
What is Reverse inference?
- From the fact that a brain region has previously been associated with a cognitive function, we infer that, when this region is activated in a different experiment, the same cognitive function is involved.
- However, is that really justified?
- Example of Hensen et al., 2000: dorsal left premotor cortex is not just involved in timing but also in many other cognitive tasks!
- Structure-to-function induction can be problematic
41
Arguments against structure-to-function induction
- Activation may be only epiphenomenal
- The function that has been assigned to a specific brain region may be incorrect or incomplete
- Activation in a certain brain region does not necessarily mean that the cognitively relevant function is carried out there (greater activity ≠ “important” or “stronger” processing)
- Two tasks/mental processes may activate a common region that has an overarching function (and thus, common activation does not imply the same cognitive process)
- The same brain region may perform two (or more) separate functions, depending on context, task, sensory input etc
- Some mental processes may require a suppression of brain regions (which may cancel out positive activations)
- Absence of activity does not imply absence of functional involvement (statistical threshold)
42
An example of when structure-to-function induction can be justified and useful (Gazzaley et al., 2005):
- A large body of evidence has identified some regions that are selective for certain processes: e.g. the fusiform face area (FFA) for face processing, the parahippocampal place area (PPA) for scene processing
- We can use these known functions to test hypotheses about cognition.
- E.g: Why do older adults show lower performance in working memory tasks?
43
What was Gazzaley et al (2005) study with faces and scenes?
- Participants given a series of faces and scenes to look at
- Either told to look at faces (and ignore scenes) or look at scenes (and ignore faces)
- Given a 9 second gap then asked whether they had seen the face/scene before, supressing what they were asked to ignore
- Testing the ability to hold information in the STM and supress irrelevant information
- FMRI results looked at brain activity in scene-selective areas (PPA)
44
What was Gazzaley et al (2005) study with faces and scenes findings?
-Brain activity in PPA or FFA shows that older adults show less suppression of distractors in a working memory task – this provides a possible explanation for their impaired working memory performance and the underlying cognitive impairment
- When the functional specialisation of a brain region is well established with a large body of different studies (preferably using different neuroimaging methods), then it may be justified to infer functional involvement from the activation of that brain region.
- However, structure-to-function inductions need to be carried out with extreme care!
45
Which kind of inference can we draw from functional neuroimaging?
- Function-to-structure deduction refers to the use of qualitatively different patterns of activity over the brain
- Allows us to distinguish between competing cognitive theories
- Structure-to-function induction is a (probabilistic) inference of a cognitive process from activation of a specific brain region
- Can help us to test theories about cognitive functions
46
Assumption of modularity of the brain:
-Pure insertion, function-to-structure deduction and structure-to-function induction all make the assumption that different cognitive functions are carried out by separable cognitive modules that are implemented as functionally specialised areas in the brain.
47
How can different information be registered in different cortexs?
- Auditory information in visual cortex (Vetter et al., 2014)
- Visual information in auditory cortex (Meyer et al., 2010)
- Information from cars and birds in FFA (Gauthier et al., 2000)
48
Can neuroimaging predict cognitive theories?
- Cognitive theories don’t always make predictions about brain responses. Therefore, it is difficult to address some cognitive theories with neuroimaging data.
- “In fMRI studies and lesion studies that investigate the functional neuroanatomy of cognition, the validity of the interpretation lives and dies with the validity of the cognitive theory on which it depends” (Wixted & Mickes, 2013)
- Not all questions in psychology can be addressed with neuroimaging data (as is true for any other type of data, too)
49
Can functional imaging ever tell us MORE than behavioural experiments?
- Sadato al. (1996) Activation of primary visual cortex in blind people when they read Braille
- This tells us that areas specialized for vision can process tactile information in the blind. This would have been impossible to find out with behavioural measures!
50
What did Owen et al (2006) find about vegetative state?
- Detecting awareness in the vegetative state
- Brain activity patterns can be used to communicate with patients in the vegetative state who are otherwise unresponsive
- Looked at a patient who had been in a car accident and was unable to communicate
- If speech was played, speech areas were activated
- Was told to imagine playing tennis or walking through their house – same brain activity areas lit up as healthy controls
- Shows conscious – able to reply to instruction
51
What did Wager & Atlas (2013) find about pain and placebos?
- Pain is a subjective experience and difficult to measure behaviourally
- How does the placebo effect work? When someone reports reduced pain, is it because they are experiencing less pain or because they have altered their decisions about that pain?
- Brain activation of pain networks can be used as a biomarker – the results suggest that the actual sensory and affective aspects of pain are reduced with placebo, pain judgment is down regulated
- Certain brain areas that light up when there is pain
52
What fMRI cannot tell us
- Causal role of a particular brain area in a particular task/cognitive process cannot be revealed – fMRI is correlational
- Low temporal resolution: fMRI does not reveal the workings and components of thought
- Despite high spatial resolution: each voxel pools neural activity from hundreds of thousands of neurons - > subsampling
- A specific neural representation for a stimulus or cognitive process does not tell us whether that representation was computed locally or whether it was inherited from a different region (Mather, Cacioppo & Kanwisher, 2013)
53
Why is neuroimaging effective in the future?
- The volume of studies has increased, we are building convergent evidence for the functional anatomy of the brain
- MRI technology and analyses techniques advance very fast
- Not so dependent on subtraction anymore
- Study of networks, connectivity, activity patterns in addition to localization
- Combining methods: fMRI + EEG -> to address questions about both when and where fMRI + TMS/tDCS -> allows us to explore the effects of interventions on brain networks
- We’re also getting better anatomical maps of the brain
- With newer scanners, functions can be more specifically localized (higher spatial resolution).
- Detailed anatomical data can help to constrain our understanding of functional networks
Conceptual
flashcards
Decks in class (10)
# Cards
