L13 - linking senses and behaviour to memory

Question 1

what are David Marr’s 3 levels of analysis

Answer 1

1) computational problem
2) algorithm
3) physical implementation

Question 2

what is the computational problem

Answer 2

integrate noisy sensory input and decide the correct action

Question 3

what is the algorithm

Answer 3

drift-diffusion model

explains - frequency of distribution of reaction lines and the speed accuracy trade off

Question 4

what is the physical implimentation

Answer 4

ramping up the neural activity to a threshold - neural mechanisms behind this not completely understood

Question 5

what is the barbarismus strategy for evidence accumulation

Answer 5

start with abstract variable at neutral point, when see matched characters will go up, mismatched will go down
will reach a decision bound stating either the messages are in depth or not in depth
y-axis = evidence
x -axis = number of characters tested

Question 6

what does it mean if get to the threshold

Answer 6

it will be 100 x more likely that is in depth/not in depth 
is log(2)

Question 7

outline the drift diffusion model

Answer 7

start at neutral point
there is an abstract variable and as evidence comes in over time, go up or down then hit a boundary to say matching or non matching
depending on how long it took to get to that boundary have a different reaction
the variable drifts up/down based on evidence coming in
the variable diffuses towards one direction bound

Question 8

what type of skew is the distribution of reaction times

Answer 8

non-Gaussian - asymmetrical - skewed to the right

Question 9

how can you transform non-gaussian data to become Gaussian

Answer 9

plot the reciprocal of the latency

so that the random underlying variable is 1/reaction time

Question 10

what is the equation for reaction time

Answer 10

t = A/r

Question 11

what is A in the chart

Answer 11

A = height of decision bound

Question 12

what it the speed accuracy trade off with mice

Answer 12

mouse pokes nose in odour, when it hears a buzz it has to stop sniffing and chose a water port based on the odour it smelled.

When the odour sampling time is cut short by the buzzer, the mouse makes less accurate decisions

Question 13

what shows why being forced too early would make decisions less accurate

Answer 13

drift-diffusion

• if cut off decision making too early, haven’t collected sufficient evidence so some may be on the wrong side.
• if accumulated more evidence would get to the correct decision

Question 14

what happens to drift diffusion if the 2 odours are similar

Answer 14

evidence is conflicting so drift up and down so the average drift to decision making bound is too slow

Question 15

what adjusts the trade off between speed and accuracy

Answer 15

moving the decision bound

Question 16

what are complications with moving the decision bound?

Answer 16

decision bounds might get lower, making the need to decide more urgent

Question 17

what area of the brain of a monkey is responsible for generating the saccades

Answer 17

the lateral intraparietal area of monkeys

Question 18

what occurs if the input is nosier ?

Answer 18

activity of the LIP neuron rises more slowly

Question 19

what occurs if have a stronger sensory input ?

Answer 19

LIP activity changes faster

Question 20

where does evidence accumulation occur in Drosophila

Answer 20

dendritic integration in Kenyon cells

Question 21

what happens in fly FoxP mutant

Answer 21

takes abnormally long to decide between two odours

Question 22

why do FoxP mutants take longer to decide

Answer 22

have excess K+ channel expression, making them leaky. This slows down their rate of evidence accumulation (depolarisation up to spike threshold)