time and causality Flashcards
1
Q
what is asociative learning for
A
- learning about causality
- learning about concurrence - what goes with what
2
Q
how does associative learning work
A
- if you had to design a learning mechanism to encode causality, what would it be like?
- you only want associations to form when there’s a real causal relationship
3
Q
what should associative learning detect
A
- directionality: cause–>effect
- sensitivity to delay between cause & effect (trace interval)
- sensitive to corr
- no learning about predictable outcomes as already known cause
4
Q
associative learning being sensitive to correlation
A
- perfect correlation - good signal
- same number of pairings but US occurs alone = bad signal
- same number of pairings but CS occurs alone = bad signal
5
Q
what possible rules are there for forming associations?
A
- would a pure contiguity model work? - hebb
- when CS & US are paired, change in associative strength proportional to intensity of CS & intensity of US
- can only explain delay
- cannot explain direction, corr, or predictable outcomes
6
Q
rescorla wagner avoiding problems from hebb
A
- allows selective learning about surprising outcomes
- when CS & US are paired, change in associative strength proportional to intensity of CS, intensity of US, and how surprising US is
if you look at a learning graph, it becomes an asymptote due to learning becoming less surprising
7
Q
how does rescorla wagner improve upon hebb issues
A
- can explain sensitivity to corr
- learning reduced through extinction
- association reduced by presenting CS alone
- learning reduced through blocking
- learning reduced by presenting US alone (importance of contextual cues)
8
Q
rescorla wagner
A
- can explain correlation & predictable outcomes
- cannot explain direction & delay
- cant explain why backward conditioning doesnt work & cannot easily explain the effect of trace intervals
9
Q
wagner’s SOP (1981)
sometimes opponent process theory
A
- incorporates time by exploting idea that processing of a stim can vary as a function of time & recent events
- stim processing is reduced if: the same stim has just been presented, a predictor (CS) for the stim has just been presented
10
Q
SOP general assumptions
A
- stim represented as a set of elements, some of which may be activated by stim presentation
- elements may be inactive, or in 1 of 2 states
- when a stim is presented, some of its inactive elements enter A1, then rapidly decay into A2 & then slowly become inactive again
11
Q
differences between A1 & 2
A
response elicited by A2 less intense than that elicited by A1 - may even oppose A1 activity
particularly important for drug tolerance
12
Q
how does SOP produce self-generated priming?
activation of food elements when food just presented
A
- first time stim presented, elements go into A1
- when food taken away, quickly decay into A2
- slowly decay into inactive again
- elements cannot go from A2 directly to A1
- only a finite number of elements
- if the next US occurs before this decay is complete, weaker response
- if associate of stim is presented, elements are activated directly to A2 state
13
Q
what happens if food is actually presented after a CS that predicts it
A
- only the few inactive elements available to enter A1
- overall response to food is weaker if it is predicted
- conditioned diminution of UR
14
Q
in order to form an excitatory association
A1 & A2
A
- CS must be A1
- US must be in A1
15
Q
in order to form an inhibitory association
A1 & A2
A
- CS must be A1
- US must be in A2
16
Q
how does conditioning happen
A
- conditioning can only happen once food has occurred
- A1/A1 overlap –> excitatory conditioning
17
Q
how does conditioning stop
A
- presenting can opener puts food elements into A2 before food is presented
- so when food actually occurs there
- is less A1 activity (conditioned diminution of UR)
- less A1/A1 overlap & less excitatory conditioning
- some A1/A2 overlap & inhibitory conditioning
- eventually they cancel each other out –> no net learning
18
Q
how does extinction happen
A
- CS in A1
- US all in A2
- –> inhibitory learning
19
Q
how does inhibitory conditioning happen?
A
- establish can opener –> food association
- sound of can opener puts food elements directly into A2
- an inhibitory prevents inactive elements of US from entering A2
- it will thus interfere with action of a conditioned excitor, which is trying to put inactive US elements into A2
- SOP predicts properties of conditioned inhibitors
20
Q
blocking
A
- early stage 1: whistle –> food so strong predictor of food
- late stage 1: food goes straight into A2, can opener where elements go into A1
- stage 2: CS mainly A1, US in A1 and A2 –> mix of excitatory & inhibitory learning
21
Q
excitatory conditioning short ISI
A
- mainly A1/A1 –> strong excitatory association
- interval between stim impact response
- if there is a A1/A1 overlap this causes excitatory learning
- can be associated more easily
22
Q
excitatory conditioning longer ISI
A
fewer CS elements in A1 by the time food is presented –> weaker excitatory association
23
Q
excitatory conditioning very long ISI
A
no CS elements in A1 by the time food is presented –> no excitatory association
24
Q
backward conditioning
A
- CS in A1, US in A1&2 –> mixture of excitatory and inhibitory learning
- compare activity in the food representations
- food then can opener
- interested in can opener predicting food
- learning starts when there’s the overlap
- forward all in A1 (A1 and A1 overlap)
- backward some in A1 some in A2 (A1 and A2 overlap)
25
applications of sop
drug addiction, tolerance & overdose
26
drug addiction, tolerance & overdose
| siegel (2001)
* tolerance occurs when addicts experience less effect for the same quantity of drug. thus they will increase their intake to achieve the same high
* many addicts die of drug overdose
* not amount of morphine in system which causes the problem
* it is not the amount of heroin that is killing them
* so what is it? - the answer lies in conditioning
27
conditioning & drug addiction & tolerance
* think of taking drugs as a conditioning trial
* drug is the US
* environmental cues that accompany drug use are the CS, and come to predict the drug
* effect of the drug is the UR; but the CR can be a compensatory response that counteracts the disturbance on the body produced by the drug
* wagner model: response to A1 state is UR, CR is response to A2 state
* this explains tolerance, with repeated experience of the drug, there are repeated pairings of the environment CS and the drug US. thus the environment will come to predict the drug, and offset the effects of the drug
* more and more is needed to obtain the same effect - effect is offset by it being predicted
28
conditioning & overdose
* what happens if you take away the environmental cues that predict the drug?
* offset of effect is gone
* don’t have compensation going on anymore
* compensatory effect of the CR is removed, & the UR increases even if the dose is the same
* many overdoses can be explained by a change of context, where the addict administers in an unusual environment
29
SOP model explain drug addiction
| 2 ways explained
* Baker & Tiffany (1985) used the model to explain how CSs predicting a drug could reduce its subsequent effects, through the mechanism of conditioned diminution of the UR
* Paletta & Wagner (1986) extended this analysis by allowing that the CR might not just reduce, but also oppose the effect of the UR
30
baker & tiffany (1985) using SOP to explain drug addiction
* retrieval-generated priming allows the CS to send drug US elements directly into A2
* fewer available to enter A1 when drug presented, resulting in a reduced CR
31
paletta & wagner (1986) using SOP to explain drug addiction
* if the CR is in the opposite direction to the UR, this could actively counteract the effect of the drug: conditioned compensatory response
* this account says the response to A1 activity may be opposite to the response to A2 activity, not just weaker
* conditioned diminution of the UR: the CS will have no effect in drug absence
* conditioned compensatory response: the CS will produce the opposite effect to the drug in the drug's absence
32
paletta & wagner (1986)
* UR to morphine is hypoactivity and analgesia, which is followed by hyperactivity and hyperalgesia
* this could mean that hypoactivity and analgesia are the A1 response, and hyperactivity and hyperalgesia the A2 response
* experiment 1: tested activity & analgesia, morphine & saline, tests at series of intervals after injection
* activity: A2 is opposite to A2
* algesia: A2 is just reduced A1
* experiment 2: tested activity in distinctive context, & then analgesia, under morphine & saline, test at series of intervals after injection
* in context, group that had experienced morphine in that context in training showed less hypoactivity both in presence of morphine, and in its absence
* in morphine test group, less analgesia in the presence of morphine but not in its absence
