Decision Making

Question 1

How can sensory inputs help you with decision making?

Answer 1

It can help you to decide if a decision will be good or bad

Question 2

Describe the perceptual decision making test you can do in a lab with dots?

Answer 2

You move / flash dots and get a person to decide whats happening:

Make dots more left for instance = easy to DECIDE which way dots are moving

You can make some of the dots flicker and test movement = harder decision

You can have almost all the dots flicker and some to move left = much harder.

Question 3

What are david Marr’s 3 levels of analysis?

This helps us to study biological problems and decision making

In order to understand perceptual decision making mainly:

Answer 3

Question one - whats the computational problem
- what are you trying to solve? E.g. deciding which way a dot is moving

Question 2 - what is the algorithm

• what does the sensory information you’re given tell you?
• some dots are flashing and some are moving

Question 3?

• what physical implementation should be taken?
• in reference to dot experiment you could increase neuronal activity to help us to better see which way the dots are moving

Question 4

What is the drift diffusion model of evidence accumulation?

Answer 4

This relates sensory inputs to whether they match a boundary or not

Essentially if you have enough sensory input which meets a boundary (as in enough evidence to tell you something is fact) then you can a decision.

Whereas if you didnt have enough sensory input to tell you something is fact (then you would have a non match batch boundary)

Question 5

A typical behavioural experiment?

Answer 5

A human or monkey looks at fixed dots on computer screen

They move right or left

A camera tracks their eyes

Rt = reaction time from when a stimulus appears and how long it takes a subject to respond.

Question 6

What is weird in the distribution Of reaction times in the typical behavioural experiment with dots?

Answer 6

The data is not gaussian (doesnt make a bell curve)

We need to transform it so we can make a Gaussian graph

Question 7

How to make a Gaussian plot (a bell curved graph)? And why do we want this?

Answer 7

To make a graph guassian you plot the reciprocal of latency (divide by one). This just means 1 / reaction time

the type of question will likely ask whats wrong with a particular graph that measures decision making. The answer will probably be that the data isnt in the centre of the graph and isnt bell curved.

We want Gaussian distributions which a bell curved to find commonalities and extremes in data.

In the example of dots and reaction times will help you to find common reaction times

Question 8

What is the equation for accumulating evidence (R)?

Answer 8

R = A/T which is the same as t =A/R

Note the first equation means evidence accumulation rate (R) is equal to 1 (A) / reaction time (T) REMEMBER this!!

Also remember that to get a Gaussian distribution from this you do the reciprocal or reaction time 1/T

R = rate of accumulating evidence before a decision is made

T = reaction time

A = height of decision bound.

Question 9

Describe the reaction time in drosophila task? And what is the skew in data?

Answer 9

Put flies in a chamber
They choose between a ‘really’ bad odour and a ‘’less’’ bad odour

In the easy DECISION making test the really bad odour is at 0.1 conc of its normal concentration. This is easy to detect as the really bad odour is fairly strong. This is a QUICK response and the flies choose the less bad odour

In the hard task the really bad odour is at 0.9 concentration - flies makes more mistakes and hesitate to decide if the odour is the really bad odour and how to respond to it. They take a much LONGER time

The harder the odour test is, there is a right skew (as in reaction time becomes slower)

Question 10

Describe the speed accuracy test: what can we decide from it?

Answer 10

A mouse pokes its nose in an odour port - when it hears a buzz it has to stop sniffing and choose a water port based on the odour it smelled. Rewards for the correct port.

Assume it has to go towards a smell which gives a reward.

Conclusion:

The longer a mouse has to decide the more accurate of a decision it makes.. when the time it has to sample a smell is cut short by a bell the mouse makes a less accurate decision.

Thus increasing the SPEED the mouse has to detect a smell DECREASES accuracy

Question 11

What does the speed accuracy test - which tells about how speedy decision affect accuracy in mice, tells us about making decisions too early?

Answer 11

Making decisions too early and deciding on evidence too early makes you make the wrong judgements

Thus you should leave experiments for longer periods of time so MORE evidence can be accumulated so a decision can be made

Question 12

What if evidence is similar? ( as in what if two smells you have to decide between are very similar)

Answer 12

Then its harder to make a decision

Thats why long evidence accumulation is required.

Question 13

What might happen to the time it takes to hit a decision boundary when in a life of death scenario?

Answer 13

The decision bound time is smaller

Question 14

In the drift diffusion model (looking if you have enough evidence to meet a boundary to decide if something is fact) overtime?

Answer 14

Decision bound times may decrease

This may be to familiarity or urgency

Question 15

What region of a monkeys brain is responsible for sending signals for the monkey to move its eyes?

Answer 15

The visual cortex - particularly the lateral intraparietal areas and its neurones.

Question 16

What is the experiement done on monkeys to test decision making

Answer 16

Is literally the dot movement task

Monkeys eyes and neurones are recorded to see when a monkey detects dots and decides on the direction they are moving.

Question 17

What can be concluded from the monkey dot experiment?

Answer 17

When movement stops of the dots, decision making especially in the LIP neurones in the visual cortex (lateral intraparietal area)

Stronger sensory input (as in less noisy input - so only dots are moving) LIP activity increases more quickly

More noise sensory input (as in dots flash and move) motion is more difficult to detect and LIP activity is SLOWER to make a decision.

LIP neurones reach a common threshold before a decision is made (they hit a boundary and then a decision is made)

Question 18

What is the link between evidence accumulation and the LIP neurone?

Answer 18

When a LIP neurone receives enough information thats enough to trigger an action and the eye moves in a particular direction.

Question 19

What do kenyon cells in drosophila do? Whats wrong with a FoxP mutant? How does this affect evidence accumulation?

Answer 19

Kenyon cells store olfactory memorys

A fly mutant FoxP takes abnormally long time to decide between two odours

This is because the FOX p mutant of kenyon cells have excess potassium channels expressed in their membrane, this makes them leaky. Potassium channels work to hyperpolarise a cell membrane. This pushes the cells to potential which is negative.

This decreases evidence accumulation - remember a depolarisation threshold is needed for a action to happen, which now takes longer due to the membrane being hyperpolarised.

Question 20

What are kenyon cells connected to? And what happens to connections of kenyon cells with optic neurones?

Answer 20

Kenyon cells are connected to output neurones

These connections change when we accumulate a memory or learn something

Question 21

Issue of leaky potassium channels?

Answer 21

High influx of sodium is needed for depolarisation to occur

Question 22

What the mutations in kenyon cells tells us about decision making? And what does monkey experiments with dots and LIP neurones tell us?

Answer 22

Mutation in kenyon cells suggest evidence accumulation and decision making is due to a threshold membrane potential being overcome - allowing for depolarisation

WHEREAS - recordings from monkeys suggest decision making is due to the RATE that neurones work like LIP neurones