Brainscape's Knowledge GenomeTM

Browse over 1 million classes created by top students, professors, publishers, and experts.


See full index

Biological Basis of Behaviour II > Discrimination & Generalisation > Flashcards

Discrimination & Generalisation Flashcards

1
Q

KEY PHENOMENA

A

GENERALISATION
GENERALISATION DECREMENT
PEAK SHIFT
TRANSPOSITION
TRANSFER ALONG CONTUNUUM (TAC)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

GENERALISATION

A
  • some response occurs to stimuli physically similar to S+ BUT not identical
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

GENERALISATION DECREMENT

A
  • response to other stimuli = < to S+ itself
  • generalisation gradient = graph relating generalised responding to values on stimulus dimension; sharpens when S- introduced
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

PEAK SHIFT

A
  • responding may be greater to stimulus other than S+ (S’); on other S+ side from S- on stimulus dimension
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

TRANSPOSITION

A
  • if discrimination between S+/S- = trained -> S’ = tested VS S+/S- may be chosen
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

HANSON (1959)

A
  • used peak-shift
  • function for controls = centred on S+ (stimulus paired w/food); no S- was used for this group
  • other curve is for animals that also had S-
  • function peak shifted from S+ away from S- aka. peak-shift
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

PEAK SHIFT

A
  • occurs as dif between generalisation gradients for excitation (+)/inhibition (-) > at point on dimension just past S+ in direction away from S-; falls off as one moves further along dimension
  • works for humans too (Livesey & McLaren)
  • prediction = peak shift works best w/similar (close on dimensions) S+/S-; shift is greatest here
  • discrimination = difficult BUT not impossible
  • not that they can’t physically distinguish between stimuli BUT generalise between them lots
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

WILL & MACKINTOSH (1998)

A
  • Q = how are stimuli represented on dimension?
  • artificial dimension created via dif icons chosen systematically as shown according to dimension position; used pigeons
  • certain features/representational elements drop out as you move along dimension; others come in
  • aka. each stimulus shares many elements w/neighbours BUT fewer w/more distant stimuli
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

WILL & MACKINTOSH (1998): RESULTS

A
  • good peak-shift could be obtained w/artificial dimension constructed like this
  • suggests that artificial dimension was behaving like “natural” one
  • provides evidence for generalisation account of peak shift
  • humans also show it; could have consequences for choice behaviour (ie. art appreciation/facial attractiveness in mate choice)
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

CLASSIC THEORETICAL ISSUE

A

ABSOLUTE VS RELATIVE DISCRIMINATION
- animal learns to respond to one stimulus over another in any discrimination
- but what is the effective stimulus (absolute/relative)? ie. does rat learn to respond to black > white/darker > lighter?
- these become embroiled in 2 complete perspectives on discrimination learning; 1 derived from early behaviourism, other from Gestalt ideas sympathetic to cognitive interpretations
- supposed crucial experiment = transposition of discrimination to dif values on stimulus dimension
- explanation of transposition based on rules (relative) VS generalisation (absolute) at stake

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

TRANSPOSITION: WILLS & MACKINTOSH (1999)

A
  • pigeons trained on S+/S-; learn to choose S+/ignore S-
  • tested; offer “near” discrimination; see which stimulus they choose
  • tend NOT to choose S+ (transposition)
  • consistent w/choosing darker stimulus; evidence of animals using rule?
  • key test = “far” discrimination; should also choose darker stimulus at same rate as before
  • BUT they don’t; roughly at chance when discrimination = successive (stimuli shown one at a time); rule forbids this so why?
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

WILLS & MACKINTOSH (1999): TRANSPOSITION EXPLANATION

A
  • in terms of discrimination on basis of absolute values; hence mere existence of transposition doesn’t establish relational learning in animals w/provision that it must be expected to reverse at extreme values on dimension
  • aka. simple another pea-k-shift! animals prefer darker stimulus + S+ as that’s where peak is
  • BUT as you move away from dimension, peak = passed -> darker pref drops
  • should be strong pref for dark stimulus VS S+ in near discrimination BUT 50/50 in far
  • absolute discrimination account based on generalisation fits facts
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

TRANSFER ALONG CONTINUUM

A
  • pre-training on easy problem followed by shift to hard problem can be more effective > training on only hard problem (even when total training times are equated)
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

TRANSFER ALONG CONTINUUM: LAWRENCE (1952)

A
  • first reported w/rats
  • 2 groups trained; 1 = hard problem (similar stimuli); 2 = easy problem
  • switch easy rats onto hard problem post training; hard animals just continue
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

LAWRENCE (1952): RESULTS

A
  • easy group learns BOTH problems quicker despite having same total trials & < trials on harder problem than “hard” rats
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

LAWRENCE (1952): EXPLANATION

A
  • training on easy problem (E+ VS E-) exploits bigger dif between curves for issue
  • training on hard problem gives hardly any dif between H+/H- (aka. lots of generalisation); this is why it’s difficult
  • humans also show TAC effect; could have implications for training phoneme perception (ie. training foreign speakers in English via giving easier problems before transferring to actual learning)
17
Q

KRECHEVSKY & LASHLEY’S NON-CONTINUITY THEORIES

A
  • discrimination of relative stimuli
  • non-continuity theories = learning occurs suddenly
    KRECHEVSKY (1932)
  • rats form hypotheses about what’s to be discriminated; when hypothesis = correct -> problems instantly solved
  • predicts position habits; no impact of pre-solution reversal/transfer along continuum
  • aka. rule-based system for learning; doesn’t work for many phenomena
18
Q

HULL-SPENCE CONTINUITY THEORY (1936)

A
  • associative approach; v successful
  • discrimination of absolute stimuli
  • continuity theory = learning occurs gradually
  • assumes smooth generalisation gradient around stimuli to which training actually occurred
  • assumes excitatory generalisation around S+/inhibitory around S- (hypothetical internal response tendencies)
  • observed response tendency predicted from unspecified monotonic transformation of algebraic sum of excitatory/inhibitory generalised response tendencies (ie. excitatory - inhibitory)
  • w/appropriate shape choices for 2 gradients -> transposition/peak-shift/transfer along continuum predictions
19
Q

COMPROMISE THEORIES: COMBINING CONTINUITY/NON-CONTINUITY THEORY

A
  • discrimination involves both learning what stimulus dimensions to attend to/what stimulus values on dimension = correct
    SUTHERLAND & MACKINTOSH (1971)
  • attentional learning = slower to reach asymptote > response learning
  • allowed attention to multiple stimuli BUT assumes attention = limited so increased attention to 1 dimension -> less to another
  • predicts overtraining reversal effect/overtraining impact on relative intra/extra-dimensional shift
20
Q

DEBATE RESOLUTION

A
  • associative theories = correct
  • BUT need ^ control of processes (ie. attention) to explain all phenomena
  • standard modern position

Biological Basis of Behaviour II (25 decks)

Brainscape helps you reach your goals faster, through stronger study habits.
© 2024 Bold Learning Solutions. Terms and Conditions