NEUR1020 Quiz 5 Content

What is learning? What are the 2 main forms of learning?

Learning= the modification of behaviour through experience

These 2 main forms involve simple changes in the strength of a pre-existing behaviour with no new behaviours emerging or existing ones disappearing:

1. Habituation (Weakening behaviour)

   • the reducing intensity of a reflex, weakened through repeated exposure and experience dealing with the stimuli

   • eg. Aplysia Californica has a gill withdrawal reflex where the gill retracts when disturbed. The more a disturbance occurs, the less the gill will retract in response to it as it gets used to it.

2. Sensitization (Strengthening behaviour)

   • the withdrawal response can be strengthened in response to the same stimulus

   • eg. the fish’s reflex can be heightened if a new, more powerful disturbance stimulus occurs. Now later, normalised disturbances will be withdrawn against more strongly than before (better safe than sorry ideology)

Habituation and sensation are both highly adaptive ways of allocating/prioritising different behaviours so that only the needed amount of response to a stimulus is generated as to not drain resources

What is classical conditioning? Describe the process. What is higher order conditioning?

When a previously neutral stimulus elicits the same response as another stimulus after repeatedly being paired together

• it is a process of learning associations between stimuli by pairing them together, eg. when a dog starts drooling when they hear their owner’s footsteps toward their food bowl.

• these associations are learnt with time and repeated experience

An unconditioned stimulus (eg. dog food) has an unconditioned response (eg. salivation) that is automatic. A conditioned stimulus (eg. footsteps) will after repeated pairings with the unconditioned stimulus bring about the same response, called the conditioned response (eg. salivation). This response is learned.

• over time, the conditioned stimulus can create the conditioned response without the unconditioned stimulus (salivating when they hear footsteps without there being dog food)

Higher order conditioning:

• pairing a conditioned stimulus (Eg. footsteps) with another stimulus (eg. a tennis ball) to create an association between the new stimulus and the conditioned response (eg. salivating when the dog sees a tennis ball)

• conditioning responses by higher-order conditioning tends to be weaker the further removed a new conditioned stimulus is from the original

What is operant conditioning? Define Thorndike’s Law of Effect.

Learning whether our actions are followed by satisfying or unsatisfying consequences

• Thorndikes law of effect= behaviours that are followed by satisfying outcomes tend to be repeated, while behaviours that are followed by unsatisfying outcomes tend not to be

• reinforcement= outcome that increases the strength/frequency/probability of a behaviour

• punishment= outcome that decreases the strength/frequency/probability of a behaviour

• what makes an outcome reinforcing or punishing depends on what has been added (positive) or removed (negative) from the environment

  • positive reinforcement= something pleasant added

  • negative reinforcement=something unpleasant removed

  • negative punishment= something pleasant removed

  • positive punishment=something unpleasant added

What are schedules of reinforcement? What are the 2 classes of reinforcement schedules?

The consistency with which a particular behaviour elicits a particular consequence

1. Continuous reinforcement= every instance of the behaviour is reinforced

2. Partial reinforcement= only some instances of the behaviour are reinforced

   • tends to produce more robust learning and continuous schedules

There are 2 types of partial reinforcement schedules that can be fixed (reinforcement provided regularly) or variable (reinforcement is irregular):

• Interval schedules= time-based reinforcement

  • fixed interval schedules= reinforcement is only delivered if the target behaviour occurs within a specific time interval (no matter how many times the behaviour occurs in that interval the reinforcement is the same)

  • variable interval schedules= each time interval is different. Reinforcement still occurs every time the behaviour occurs in the interval but the end of the interval/arrival of reinforcement is harder to predict.

• Ratio schedules= response-based reinforcement

  • fixed ratio schedules= setting a behaviour ratio, such as 2:1, and providing reinforcement every second instance of the behaviour regardless of how much time has passed between each occurrence

  • variable ratio schedules= the number of responses required to elicit reinforcement varies unpredictably

Generally, variable schedules produce steady response rates while fixed schedules produce more eratic ‘stop-start’ patterns of responding

Describe the Multi-store model of memory (aka Modal Memory)

Encoding, storage and retrieval all form parts of the interactions between 3 memory components.

1. Sensory memory

   • immediate perceptual information is stored in separate streams

   • each of our perceptual modalities (vision, olfaction, taste, touch) has its own dedicated memory store

   • sensory information must be attended to and transferred to short term memory to avoid being forgotten within a brief time frame

     • as we are bombarded with far more sensory information than we could possibly remember, most sensory memory is lost to forgetting

     • all separate streams are combined in short term memory (it isn’t modality specific)

     • the time frame varies to the sensory store: iconic memory (visual) can encode 12+ distinct visual objects at a time but lasts <1s, and echoic memory (auditory) can only encode 5 auditory objects at a time but lasts 5-10s.

2. Short Term memory

   • houses the current contents of awareness

   • information here is available for report

   • has a capacity of 7+-2 items of awareness at once before performance suffers

   • information here can be refreshed through rehearsal (mentally repeating information) to stop it from being lost to forgetting.

     • rehearsal allows info to persist 20-30s

     • can be achieved sub vocally (without speaking aloud)

   • rehearsal facilitates the transfer of information to long term memory: any information retained for longer than 30 seconds is stored in long term memory

3. Long term memory

   • stores memories that can be retrieved into short term memory to be repeated or used to support ongoing processing

   • STM and LTM freely pass information via transfer (STM→LTM) and retrieval (LTM→STM)

   • retrieval to STM is often prompted by cues available in STM

   • LTM has massive, perhaps limited, capacity. Once encoded into LTM retention is permanent: failures to remember long term information are due to failures of retrievel, not failures of storage.

What are the 2 theories for why short term memory is forgotten?

1. Time-based decay (aka temporal decay)

   • information in memory is progressively degraded by the simple passage of time (when not actively maintained or refreshed through rehearsal)

   • extending the retention interval (time period between stimulus presentation and report) will produce poorer memory performance

2. Interference

   • “distractor items” make it harder to maintain short term memory and reduce memory performance

   • the more distracting information there is, the harder it is to pinpoint and return target information

   • buildup of interference produces poorer memory performances

What is the free recall task and serial position curve? What are the serial position curve effects and how do these align with the multi-store model of memory theory?

A list of items/stimuli are presented and participants are asked to recall as many items as they can in any order

• their responses from many lists can be used to measure recall accuracy. Plotting recall accuracy vs the item’s given position in the list produces a serial position curve

Serial position curve effects:

1. Recency effect=end of list items are remembered better

   • these items are freshly stored in short term memory and therefore easily reportable

2. Primacy effect=beginning of list items are remembered better than middle (but not as much as end)

   • in order to make room for new items as the list progresses, the first items that are well rehearsed are consolidated into long term memory

   • time between the presentation of each item gives individuals an opportunity for rehearsal

   • these items can be retrieved from LTM for recall

3. Middle items have lowest accuracy that bottoms out and stabilises

   • these items had to be rehearsed alongside lots of items, reducing rehearsal effort and making them less likely to end up in LTM. Instead, they are just forgotten

What are cues?

Cues= any features of the environment that direct us toward information in memory that we are trying to access or recall

• they become associated with target information and boost the likelihood of successful retrieval, as remembering a cue can trigger memory for the target information

• More cues=greater buffer against retrieval failure

What is the encoding specificity principle?

Retrieval of information from LTM to STM is facilitated when cues that are present at encoding are also present at retrieval

• in this way, incidental features of the environment (and our mental state) can become associated with target information

• eg. I associate the song ‘Perfume’ to Chemistry because I was doing a chemistry assignment the first time I listened to it. So doing chemistry homework makes me think of Perfume, even though the 2 are unrelated in concept.

Cues: deeper encoding effect

Stimuli that are processed in greater depth at encoding have a memory benefit over stimuli that are processed more superficially

• this deeper encoding links the information with more related concepts than superficial encoding, creating more links to facilitate retrieval

Cue to action: describe the interference effect. What are the 2 types of interference?

Cues can become linked to retrieving automatic behaviours (eg. seeing a fork in the road and turning on the indicator)

• If an aspect of the associated environment changes (interference), cues that used to signal appropriate responses can signal inappropriate responses, resulting in error.

• eg. get a new car where indicator and windshield wipers have swapped sides. at a fork in the road you accidentally turn on the windshield wipers instead of the indicator due to the automatic behaviour.

There are 2 types of interference:

1. Proactive interference= older information disrupts the retrieval of more recent information (eg. indicator situation above)

2. Retroactive interference=more recent information disrupts the retrieval of older information (eg. studying Italian and accidentally saying ‘ciao’ instead of ‘hello’ in a normal English conversation)

What are the 2 ways that problems can be solved?

1. Incrementally (via a step by step process)

   • exhibited by animals: eg. Thorndike’s cats incrementally learned how to escape a puzzle box)

2. Instantaneously (by insight: an ‘aha!’ moment where the solution comes all at once in it’s entirety)

   • exhibited by animals: eg. Kohler’s chimpanzees tried to reach bananas but couldn’t then had an ‘aha!’ moment and stacked up crates to climb up and get them.

Describe the computational perspective on Problem Solving and its 2 approaches

Describes 2 different approaches to incremental problem solving:

1. Algorithms= fixed procedures for producing a solution to a problem

   • pros: always produces a solution

   • cons: can be inefficient

   • eg. building a lego house from the instruction booklet. May take a while but the exact house is guaranteed

2. Heuristics= “rules of thumb”

   • pros: require less information than algorithms, are easier to use and are quicker

   • cons: can produce sub-optimal results and a solution isn’t guaranteed

   • eg. building a lego house off of the rule of thumb that it has 4 walls and a roof; improvising the rest. Fast but the product may not exactly match what was depicted on the box.

When choosing between a heuristic and algorithmic solution, the risk of a poor outcome must be weighed against the time taken to produce a solution.

Heuristic approaches: describe Means-ends analysis vs Hill-climbing

Means-ends-analysis= breaking down an end goal into a set of smaller goals

• breaking down the problem into subpronlems simplifies the computational burden

• requires the end goal to be known

• eg. to build a lego house, first make 4 walls, then a roof, then attach them

Hill Climbing= Repeating the action that gives you the most improvement toward the goal state until progress is no longer possible. Point of inability to further improve=the solution.

• don’t need to know the end goal, just an idea of what ‘progress’ is

• eg. clibing a hill and can’t see the top. Progress=moving upward, choosing the path that most increases altitude until you reach the top/as high as possible.

• pros: highly versatile and good when the problem structure isn’t well known

• cons: can get ‘stucl’ in an intermediate state that is better than the initial but does match the goal (eg. the hill path doesn’t reach the top). This is called reaching a local maximum

What is ‘utility’ in terms of decision making? Describe the Subjective Expected Utility Theory vs the Elimination by Aspects strategy

Utility= the concept of benefit in decision making.

• decision are made to maximise the utility of our choices: the single best choice conders the maximum net utility

The Subjective Expected Utility Theory says we ought to make choices by considering: 1-possible outcomes, 2-the probabilities of those outcomes and 3-the value attached to each outcome

• A decision has certain values (eg. choosing a house: you want one that is low cost, near work, in a good area, spacious etc) and these values can be given number values for each possibility based on how much it reaches those values. By adding up the total ‘value points’ of each possibility, the best one can be identified

• rational decision making under this framework can be hard to manage and effortful.

Elimination by aspects, instead of simultaneously considering each possibility and it’s value, considers individual values one at a time and uses this to rule out possibilities until you are down to one best option

• eg. house example: gets rid of the most expensive house, then the worst area house, then the least spatious house and keeps going until you have one best option or at least a more manageable subset.

• simple, efficient, minimises information processing load

• more like how actual day to day decisions are made

What is the Gaze Heuristic?

When you are running to catch a ball you don’t calculate it’s flight trajectory to position yourself. Catching is instead accomplished by fixing your eyes on the ball, beginning to run and adjusting running speed to keep the gaze angle constant

• instead of thinking about a bunch of variables, you focus on the one variable: angle of gaze