learning via instructions at the liplab. 0. theoretical background learning = effect of regularities...
TRANSCRIPT
Learning via Instructions at the LIPlab
0. Theoretical Background
Learning = effect of regularities in environment on behavior
(De Houwer et al., 2013, PB&R)
=> three types of learning effects
- regularity in presence of one stimulus (e.g., habituation, ME)
- regularity in presence of two stimuli (classical conditioning)
- regularity in presence of behavior and stimulus (operant cond.)
Instructions – ACBS Minneapolis – 20 June 2014
2. As an effect, a specific type of learning can be due to several types of mental processes: E.g., classical conditioning
a) Association formation models
=> relatively passive, stimulus-driven formation of associations
b) Propositional models (De Houwer, 2009; Mitchell et al., 2009)
=> specifies TYPE of relation + have truth value=> active problem solving: How are events related in the world?
Pairings(e.g., tone-shock)
Change in behavior(e.g., increase skin conductance)
Pairings(e.g., tone-shock)
Propositions Change in behavior(e.g., increase skin conductance)
Associations
Instructions – ACBS Minneapolis – 20 June 2014
3. Theoretical freedom allows for new predictions
=> We focus on one prediction of propositional models about learning via instructions: The way in which the proposition is formed should not matter (if the content of the proposition is exactly the same)
=> Forming a proposition about the regularity in the environment (e.g., pairing of stimuli) via experience, instruction, or inference should be equivalent (if this leads to equivalent propositions)
=> Aim: to compare learning via experience and via instruction
- What is unique about experience?
- Can instructions be changed to mimic this unique impact?
=> To increase chances of finding unique aspects of experience, we examine types of learning that are assumed to be “low level” (i.e., mere exposure, evaluative conditioning, fear conditioning, automatic SRC effects, approach / avoidance responses).
Instructions – ACBS Minneapolis – 20 June 2014
I. Non-associative learning via instruction
Mertens et al. (in preparation)
- Mere exposure: Regularity in presence of stimuli (i.e., how often / probable) influences liking
- Instructions: You will see “Bayram” often and “Enanwal” only now and then
- Implicit measure of liking: Implicit Association Test - Multiple studies: Preference for Bayram over Enanwal on ratings
and IAT, not on evaluative priming
Instructions – ACBS Minneapolis – 20 June 2014
II. Classical conditioning via instructions
II.1. De Houwer (2006, L&M): Instructed Evaluative Conditioning
- Evaluative conditioning: changes in liking due to pairings- What if pairings are merely instructed?: If “Bayram”, then you will
see a positive picture; If “Enanwal” then you will see a negative picture.
- Implicit measure of liking: Implicit Association Test- Result: (Implicit) preference for Bayram over Enanwal even if no
pairings were actually presented.
Instructions – ACBS Minneapolis – 20 June 2014
II.2. Gast et al. (2013, L&M)
- Exp. 1:
=> Instructed EC (e.g., product 1=> pos; product 2 => neg)
=> Instructed extinction (nonwords without pictures) or counterconditioning (e.g., product 2=> neg; product 1 => pos) or control (no info about a second phase)
=> measure: evaluative ratings
=> EC in control but not in extinction or counterconditioning
- Exp 2
=> same as Exp 1 but first phase (pairings) instructed or experienced + IAT measure
=> regardless of type of first phase, EC in control and extinction but not in counterconditioning
Instructions – ACBS Minneapolis – 20 June 2014
II.3. Fear conditioning via instructions (Raes et al., 2014, PLoSOne)
Instructions – ACBS Minneapolis – 20 June 2014
Instructions – ACBS Minneapolis – 20 June 2014
Instructions – ACBS Minneapolis – 20 June 2014
Additional studies:
- modulation of startle response as DV
- replication using fMRI
- selective learning (i.e., “preparedness”)
- instructions about context-dependent relations
…
=> always compare experience-based with instruction-based FC
III. Operant Conditioning via Instructions
III.1. De Houwer et al. (2005, P&P)
- Instructions: If “left” or “<=” say BEE; if “right” or “=>” see BOO
= Sd: R – correct relation
=> but never executed- Measure: Spatial Simon task
=> if blue square say BEE; if green square say BOO
=> irrelevant square location left or right- Result: faster if irrelevant square location and location linked to
response match (e.g., say BEE for square on left)
III.2. Liefooghe et al. (2012, JEP:LMC; 2013, PB&R)
III.4. Van Dessel et al. (in preparation)
- Operant EC: like things you approach; dislike things you avoid
- “Approach Bayram – Avoid Enanwal”
=> changes implicit (IAT) and explicit liking
- “Approach Blacks – Avoid Whites”
=> does not change liking
=> actual approach-avoid does change liking (but only if participants are demand aware)
What is extra value of practice?
IV. Conclusion
We now have a wide variety of paradigms to study different types of learning via instruction.
Possible benefits:
- learn more about learning via instructions
- learn more about unique impact of experience
- learn more about how to optimize instructions
- constrain (but not differentiate) cognitive models of learning
Instructions – ACBS Minneapolis – 20 June 2014
Mutual supportive nature of functional and cognitive approach
1. What can the functional approach offer?:- mental free way of talking about instructions
=> maximizes freedom of cognitive models- RFT: instructions as event that evokes AARR
=> prediction on the basis of analogy
2. What can cognitive approach offer?- propositional models currently add little beyond relational but
more complex models can be developed, in part on evidence generated by research in functional tradition
- theories of language and reasoning
Instructions – ACBS Minneapolis – 20 June 2014