From perception to decision II

What does TMS measure for pertubation?

global manipulation across whole area

easy to test multiple areas

Can we use a more precise pertubation and still predict neural activity?

microstimulation

What was Nichols and Newsome, J’s orientation task for macaques? (hint we’ve seen this before) What is the coherence for this task?

monkey looks at FP

motion stimulus placed in MT neuron RF

upon motion onset, monkey has to saccade onto circle in direction of motion

coherence = 80%

What is this image demonstrated for the task?

black dots represent locations of saccades across a session

grey box is ‘correct zone’ and if saccade lands in box, monkey is rewarded

What were the results from this trial at 80% coherence?

accuracy is measured by comparing visual direction vs response direction

• straight line on graph indicates that monkeys are really good at the task after training

What is orientation organization like for MT neurons?

cells is neighbourhood of electrode tip have similar tuning

organized into orientation maps

What is microstimulation and what does it cause?

small electrical stimulation (local and controlled) that increases neural activity locally 

• perturb is more targeted way

• can elicit activity in different oriented neuron

How do we measure preferred orientation?

distance to centre is measure of firing activity

dashed circle is activity without any stimulus

this show tuning to rightward motion

What occurs with microstimulation and 80% coherence?

black dots = saccades without microstimulation

pink arrow = change in saccades when mircostimulation applied

with a strong signal (80%), microstimulation biases decision BUT doesn’t shift perception completely

What occurs with 0% coherence and microstimulation?

black dots = saccades without microstimulation

blue dots = saccade with microstimulation

when there is no signal (0%) and microstimulation, it drives all dots right and decides full perceptual decision

Recap: What is the difference between microstimulation with strong signal vs. no signal?

at strong signal → microstimulation biases decision

at no signal → microstimulation drives decision

What’s the difference between TMS and microstimulation?

TMS: global manipulation

microstimulation: allows us to link the insights from the monitoring approach to perturbations at a much finer scale

Describe this graph and in particular no bias case (solid line)

positive value = evidence for right choice

negative value = evidence for left choice

in the no bias condition, when there is no evidence, subjects choose either side with p = 0.5 probability 

if strong positive evidence, they choose right choice

if strong negative evidence, they choose left choice 

What occurs in the bias condition? Does it effect the behaviour with strong evidence? What does this mean?

bias condition = shift in graph to the right

however no does not impact behaviour when there is strong evidence

you haven’t broken down perception completely

What occurs when we introduce stimulation?

subjects choose equally both left or right even for positive value of the decision variable around 20

(evidence is more right and monkey chooses equally right or left)

• they answer 50/50 when we show 20% correlation to the right

• therefore manipulation provides negative evidence

How do we quantify stimulation and change in bias?

0% correlation = p=0.5

20% correlation = p=0.5

stimulation caused -20 units of decision variable

What is this experiment? What is another similar version you can do - it was video shown in class?

1. Monkey has to hold gaze onto a fixation point (FP)

2. Stimulus appears

3. Target LED appear at one of two locations

4. Monkey has to saccade to one of these locations

5. Repeat trials with different stimulus correlations

other version the same except targets are fixed and there all the time

What task did monkeys perform in the MT microstimulation experiment?

they judged the direction of motion in RDK (left vs. right) at different levels of motion coherence

What was measured in the experiment?

monkeys’ choices were plotted as a psychometric curve showing probability of choosing one direction as a function of motion coherence

What effect did microstimulation of MT neurons have?

causes shift in curve (bias)

it biased the monkeys’ decisions towards stimulated neurons preferred direction of motion

How was the effect of microstimulation quantified?

leftward shift of behavioural curve so stimulation is equivalent to rightward motion

Example from panel A:

• No stimulation → 50% at ~0%.

• Stimulation → 50% occurs at about –6%.

That means: stimulation is equivalent to adding +6% coherence in favour of the preferred direction

What is the premotor cortex responsible for?

preparation of motor activity

Explain the reaching task for premotor cortex activation?

1. start trial by placing cursor in middle

2. 2 colour cues appear in locations 180 degrees apart

3. they disappear for delay period (memory phase)

4. one of the colours appear in centre

5. move cursor to where that colour was positioned 

6. locations of colour change across trials

What do we find when doing that task and recording premotor cortex neurons?

anticipatory activity

1. spatial cues shown

= activity

2. memory period

= increased representation of the two possible directions

3. colour cue

= increased representation of the chosen option

= inhibition of non-chosen option

4. go signal

= activity for chosen option

When we compare the representations across areas, what is activation like for M1 (motor cortex), caudal PMd and rostral PMd?

M1: representation of chosen action only after cue determining action is presented

PMd: initial representation of two possible locations until colour cue presented

rostral PMd has strongest anticipation signal

What areas are active for preparation of possible motor plans vs. motor preparation and execution?

preparation of possible motor plans (memory period): rostral PMd

motor preparation and execution (after colour cue): M1

What is an error trial in reaching task?

choosing wrong direction

e.g. colour cue says ‘go to red’

picks blue direction

What does the correct vs. error trial look like when comparing representations?

error trial: more activity for error location beforehand

• preparatory activity seems driven by eventual choice (action plan) rather than perceptual stimulus

Key conclusion about errors in this task?

Neural signals in anticipation predict the chosen action (correct or incorrect) showing that errors emerge from biased choice representations, not misperception of the stimulus.

What are two more versions of this reaching task?

version 1: 2-target task

version 2: 1-target task

1. FP

2. shows one colour in one location

3. memory phase

4. shows colours at FP

5. choose location of that colour

(no spatial ambiguity during delay/memory period OR no competition between 2 locations)

version 3: reverse task/MS task

1. FP

2. show colour at FP

3. show 2 colours at different locations

4. remember colour from 2. and choose location

no spatial info in delay phase

What was found for 1-target task in terms of representations?

anticipatory activity in one direction only

What was found for match-to-sample task in terms of representations?

anticipatory activity is above baseline but untuned after colour cue

selectivity is specific to one location after spatial cue