Learning and Memory Flashcards
what is lambda?
the maximum that can be learned
what is v?
what we already know
what is lambda - v?
the prediction error
what does it mean when the prediction error is small?
there is a small amount to be learned
where does the rescorla-wagner rule originate from?
pavlovlian conditioning, as it aims to predict an outcome
what does the rescorla-wagner rule highlight?
the importance of surprise in learning and memory, as the amount learned is proportional to the amount of surprise at the outcome
what is learning?
the acquisition of behavioural information
what is memory?
the retention of information that guides future behaviour
how does the rescorla-wagner equation change when a lot has been learned?
v becomes biggeer, so l-v becomes smaller.
this means that delta-v becomes smaller, which creates the shape of a learning curve
what does the rescorla-wagner rule suggest?
there should be no learning when an outcome is already fully predicted, as lambda and v would be equal
why is the rescorla-wagner rule used?
because the presence of stimulus x ‘blocks’ the learning of y, we need to learn whether y is predictive of the outcome, or if x just has predictive value
what does blocking interrupt?
the conditioning process, and it allows for necessary adaptive behaviour in the future
what studies did hollerman and schultz conduct in 1998?
they recorded the electrical activirty in the brains of monkeys as proof of prediction error signals
what studies did hollerman and schultz conduct in 1998?
they recorded the electrical activity in the brains of monkeys as proof of prediction error signals
in hollerman and schultz’s study, what did the monkeys learn?
they learnt that particular visual imaging led to particular events.
discrimination learning was used to present them with two simultaneous visual images, and they received a reward when the correct image was pressed
in hollerman and schultz’s study, where was brain activity recorded?
found activity in the VTA- theorised this was where prediction error signals originated.
they also found the presence of neurones which express and release the neurotransmitter dopamine
in hollerman and schultz’s study, what results did they find?
in monkeys, the area of the brain associated with reward has increased activity when something surprising happens
in hollerman and schultz’s study, what happened to the braina activity of monkeys when the process was repeated?
the neurones in the VTA began to only respond when the reward was predicted, rather than when it was recieved.
this is proof of prediction error signals in the brain to predict and adapt to expected outcomes
in hollerman and schultz’s study, what happened when a reward was predicted but no reward occured?
the dopamine neurone signals decreased in activity, showing proof of negative prediction signals.
this is because the monkey had learned too much, meaning v is bigger than lambda- prediction error becomes negative
what was theorised about prediction error signals in humans?
that they would be more evident in an unpredictable situation as opposed to a predictable situation
in berns’ 2001 study, what fMRI activity was found in the human brain?
higher activity was found in the unpredictable situation (being given juice or splashed with water at irregular intervals) compared to the predictable situation
in berns’ 2001 study, where were high levels of brain activity found?
in the nucleus accumbens
what is the nucleus accumbens?
a key part of the reward system due to its conection to the VTA
it recieves signals through the mesolimbic dopamine pathway
can prediction error signals regulate learning?
learning can occur through prediction error signals due to the consequences of ‘over-expectation’
how can ‘over-expectation’ regulate learning?
when a predicted situation does not occur, uncertainty is created which reduces expectations in order for new learning to take place
what happens when two predictors of an outcome come together, and the expected outcome does not materialise?
this causes predictions to be toned down due to the high prediction error.
leads to revaluing the original stimulus, e.g., alien, predator, alien vs predator
can prediction error signals update memories?
in 2009, takahashi et al found that over-expectation training caused a change in subsequent normal behaviour
what did takahashi’s study involve?
a learning condition where rats were presented with sugar after hearing a sound (A1 and A2), along with being presented with sugar after seeing light
what was the over-expectation training condition in takahashi’s study?
the rats were presented with sugar after receiving both the A1 sound and light
in takahashi’s study, what happened to the rats between A1 and A2?
in takahashi’s study, what happened to the rats between A1 and A2?
after being presented with no reward, the rats showed a change in memory and behaviour- as the predicted situation did not occur
in takahashi’s study, what happened to the rats when their VTA prediction error was inhibited?
their behaviour did not change
this is proof that prediction error signals can update and regulate learning and memories for subsequent behaviour
