Lecture 07 - From perception to decision I

Recap: What is the ventral pathway for?

‘what’ pathway

object recognition

What is the dorsal pathway responsible for?

‘where’ pathway

motion

What region specifically is responsible for motion?

MT

What can we use to identify brain areas related to motion perception?

PET (positron emission tomography)

monitoring approach

What is PET?

functional mapping technique

uses radioactive substances to identify and measure changes in metabolic processes

• tracking blood flow/hemodynamic response

  • using O15

What is PET more commonly used for?

less used for functional mapping

allows detecting specific metabolites in disease

What monitoring approach can we use for finding areas involved in motion? Which area was found to be active?

use PET

compare responses for moving and static stimuli

V5 or MT has been implicated in motion perception

How can we use a perturbative approach to study motion perception?

looked at lesions linked to MT

= akinetopsia

What is akinetopsia? Why is it hard to study?

inability to perceive motion

• views world in stop motion

usually due to bilateral lesions in V5

bilateral lesions in V5 often also target other visual areas and aren’t localized 

• these impact other visual tasks

What is important when linking cognitive deficits to a given brain area?

behavioural testing for specificity of symptoms

• need to run a number of control tasks

What is famous persons test?

TD could tell if a photo was of a famous person, look-alike, or unknown person

What is the overlapping figures test?

TD identified all overlapping figures except one; could then recognize it when outlined with a pen.

What is the Giuseppe Arcimboldo paintings test?

TD identified both the small details (vegetables, fruits, etc.) and the overall face composition.

What is the optic ataxia test?

TD could touch the designated finger of the test leader in the air.

What is the oculomotor apraxia test?

TD could follow the test leader’s finger movement with her eyes.

What test was done for akinetopsia?

• They presented moving stimuli (dots, lines, objects) and asked her to report the direction and velocity.

• She consistently failed to perceive continuous motion above a certain speed (18 and 13 deg/s). Instead, she described it as a series of static snapshots and she could tell you order of successive positions

For the akinetopsia testing, what are these graphs depicting?

patient was asked to judge whether she could perceive a moving spot of light at different stimulus velocities

x-axis = stimulus velocity

y-axis = number of responses

yes line = number of times the patient reported seeing motion

no line = number of times the patient reported not seeing motion

What were the results from the test both vertical and horizontal motion?

could perceive motion at low speeds (2–10°/s) but above ~13–18°/s she mostly said “no”

Where in the brain order do neurons respond to static vs. moving random dot pattern? Is this conserved across species?

response in higher visual areas in both species

• in V2/V3 and MT/V5

similar architecture in higher visual areas for human and macaque

How did we measure causal relationship in macaque for MT?

lesion in MT using neurotoxic lesion

measure motion sensitivity thresholds as a function of speed and motion strength

How does coherence of dot motion control difficulty?

0% coherence means no global motion and very difficult

50% coherence = partial

100% coherence = easiest task

What impairments in motion sensitivity result from MT lesions in macaques?

MT lesions cause severe, selective deficits in motion perception

• monkeys need much higher coherence levels in RDM especially in visual field region corresponding to the lesioned MT

other aspects of vision remain intact

What is important that experimenters do to contrast study? What did experimenters do with macaques and motion test to contrast?

need a control task to make sure that motion is specific to MT

did control task of reporting orientation of static grading

• lesion in MT did not affect this

What are the properties of single neurons in area MT?

tuned for motion orientation direction like V1

• each neuron responds maximally to motion in one direction and much less/not at all to the opposite direction

How do we create population averages for motion orientation in MT neurons?

1. record firing rates across many neurons

2. preferred direction (max firing) is set to 1

3. record other neurons and average their direction-tuning

vector averaging method

How do we measure whether an MT neuron is narrowly or widely tuned?

• neurons that are narrowly tuned have population average curve with a sharper, steeper peak around the preferred direction

• neurons that are widely tuned have population curve with broader and flatter, spreading out across directions

What is found when comparing orientation selectivity across nearby neurons? What is this similar to?

nearby neurons are tuned to the same orientation

consistent with V1 orientation maps

How is speed tuning measured in MT neurons?

present moving visual stimuli at different speeds and record each neuron’s firing rate

• neuron’s preferred speed is where firing is maximal 

• produces a non-monotonic, bell shaped curve

Do neurons have different preferred speeds?

yes!

receptive field is tuned to a preferred speed

How can we create a speed relative to optimum graph?

for each neuron: divide its firing rate by max firing rate (preferred speed = 1)

plot speed as a ratio to the neuron’s optimum

• if neuron prefers 16deg/s then 8deg/s is 0.5x optimum

• if a neuron prefers 16deg/s then 32deg/s is 2x optimum

use fraction of speed relative to optimum to average neurons together to get a population picture

Recap: What are the two properties of speed tuning in MT?

1. non-homogenous distribution of speed tuning across the population

• not all neurons are tuned to the same speed

2. cells closer to each other tend to have more similar speed tuning

Describe the monitoring behavioural study on macaque to test if MT neurons contribute to perception?

1. monkey has to hold gaze on FP

2. stimulus appears (RDM)

3. two target LEDs appear at separate locations

4. monkey makes a saccade to the target corresponding to the perceived direction of motion

5. repeat trials with different stimulus correlations

What do we use a psychometric curve to measure? How do we vary the difficulty of the macaques task?

measures behavioural accuracy

vary difficulty of the decision by changing the correlation of the dots

What does this task look like graphically for an experimenter vs. a subject?

• The experimenter knows the true stimulus direction (left vs. right).

• They set the stimulus strength

• a strong stimulus → consistent correct choice.

Subject’s perspective

• The subject (monkey) doesn’t see “correlation %,” they just perceive noisy evidence.

• Their brain sets a decision boundary along a decision axis

• If perceived evidence falls to the left → choose left. If to the right → choose right.

What is noise structure in a study?

variability in neural/psychological processing → sometimes there are errors, even with strong signals.

Explain constructing a psychometric curve:

plot probability of choosing correct against stimulus strength

• creates sigmoidal psychometric curve

  • P = right choice

  • at neutral stimulus (0 direction) monkey is 50/50 left vs. right

  • as stimulus strength grows, responses get more consistent with 1 or 0

What is the subject’s reward graph?

P = reward

if monkey chooses correctly, reward probability is high (around 1)

if evidence is ambiguous, reward probability drops (around 0.5)

On the graph, what is higher correlation of dots correlated with? What is 0% correlation of dots equal to?

greater correlation of dots = greater proportion of correct answers

0% correlation of dots = 50% accuracy

What are two equivalent ways to plot psychometric curves?

1. accuracy vs. stimulus strength

2. proportion of choosing one option vs. signed stimulus strength

Describe Psychometric curve (proportion of choice)

• X-axis: Signed evidence (negative = leftward motion, positive = rightward motion).

• Y-axis: Probability of choosing ‘right’.

• At 0 (ambiguous stimulus), subject is around 50/50.

• Strong leftward evidence → probability of right choice goes down toward 0.

• Strong rightward evidence → probability of right choice goes up toward 1.

• The curve is sigmoidal (S-shaped) and centered on the decision boundary.

Psychometric curve: change in sensitivity

change in slope of curve

steeper slope = smaller noise and better discrimination (less correlation to make accurate decision)

shallower slope = more noise/less sensitivity

Psychometric curve: Change in bias

curve shifts left or right

• subject favours one choice

How does neuron’s firing rate compare to behaviour for motion discrimination in each direction?

at 0% coherence, firing looks similar for both directions

as coherence increases, firing distributions separate and its easier to predict which way the stimulus is moving from neuron spikes

What is the neurometric curve?

plots how accurately stimulus direction can be predicted based only on a neuron’s firing rate

What is psychometric vs. neurometric curve? What does comparing them tell us?

psychomtric shows behavioural accuracy

neurometric shows a single neuron’s accuracy

comparing them reveals whether that neuron could explain the subject’s decisions

What was found when study compared the neurometric and psychometric curves?

neuron is as sensitive as monkey

for some neurons, there is basically no difference between behaviour and neural activity

• some neurons are more sensitive than behaviour and some as less sensitive than behaviour

What is the average threshold ratio of neuron to behaviour? What does this tell us correlationally?

threshold ratio is centered around 1

• threshold of neuron is similar to threshold of behaviour

strong correlation of how accurate cells/neurons are determining stimulus and how accurate monkey is at behavioural task

What is correlation level of tuning of MT and behavioural ability to discriminate motion?

strong!

How can we test for perceptual sensitivity to motion discrimination? Describe task.

• Subjects are shown two gratings (moving bars) one after the other.

  • One is the reference grating (fixed at 8°/s).

  • The other is the test grating (varied speeds: faster or slower).

• Subjects report which grating moved faster.

• build psychometric curve

What is TMS?

strong magnetic field over scalp

• perturbs activity in region temporarily

non-invasive

poor spatial resolution; good temporal

can only activate superficial areas of brain

What happen when experimenters used TMS on V5/MT for static spatial frequency task?

no effect

these regions (V5/MT) aren’t critical for static spatial judgement

What happen when experimenters used TMS on V1 for motion speed discrimination?

no effect

What happen when experimenters used TMS on V1 for motion speed discrimination?

clear deficit

psychometric curve shifts (higher threshold for detection)

• no slope change since sensitivity itself isn’t worse but subjects need more evidence to make same judgement

What did the TMS motion study tell us?

casual link: V5/MT and V3a are specifically needed for motion perception