Download presentation
Presentation is loading. Please wait.
Published byCameron Chandler Modified over 8 years ago
1
Counterexample Retrieval and Inhibition During Reasoning Direct evidence from memory probing Wim De Neys Lab Experimental Psychology Leuven, BELGIUM
2
Example
3
Conditional reasoning: If I water the plants, then they grow well. I do not water the plants. Therefore, the plants will not grow wel. If P, then Q NOT P ---------------------------------- Therefore, NOT Q “ DA inference”: ERROR!
4
Searching counterexamples If I water the plants, then they grow well. I do not water the plants. ------------------------------------------ DA: The plants will wilt. 2. Searching stored counterexamples “Is it possible that the plants grow well if I do not water them?” 4. Search successful? RAIN! NO DA ??? DA! 3. Memory stored alternative causes … PLANTS GROW WATER fertilizer RAIN … cactus
5
Counterexamples and logical bias Nuclear power plant: If the alarm sounds, then the reactor is overheated. The alarm sounds. Therefore, the reactor is overheated. If P, then Q P ---------------------------- Therefore, Q! “MP inference”: VALID!
6
Retrieving disablers If the alarm sounds, then the reactor is overheated. The alarm sounds. ------------------------------------------ MP : The reactor is overheated! 2. Searching for stored counterexamples “Is it possible that the alarm sounds but there is no overheating?” 4. Search successful? EXERCISE! NO MP! 3. Memory stored disablers ALARM HEAT error EXERCISE …
7
Selective disabler inhibition Some people reason correctly –Selective disabler inhibition Markovits & Doyon (2004) De Neys et al. (2005) Handley et al. (2004) … Markovits & Doyon (2004) –Measure of inhibitory capacity “A husband loves his …”, “A teacher works in a …”, “Dogs hate …” –Conditional reasoning test Better inhibition, more MP/MT acceptance ALARM HEAT error EXERCISE …
8
Popular view –Alternatives– Disablers AC/DA MP/MT – Search – Inhibition Conditional Reasoning = Counterexample retrieval and inhibition interplay E.g., Markovits & Barrouillet (2002), Simoneau & Markovits (2003), De Neys et al. (2005) “If …, then???”
9
Explanatory burden? Memory mechanisms are crucial Only indirect evidence –Pretest more alternatives less AC good inhibition more MP CE really activated/inhibited during reasoning? –Need for DIRECT memory probing
10
Alternative accounts Classic LOGISTIC view –Reasoning is purely formal Accept inferences because of match/mismatch with rules Better pretest retrieval = smarter = better logical database = better performance … “Counterexamples play no role!”
11
Alternative accounts Classic LOGISTIC view –Reasoning is purely formal Accept inferences because of match/mismatch with rules Better pretest retrieval = smarter = better logical database = better performance … FREQUENCY view –Frequency of exceptions vs. counterexample search People do not look for “explanations” Fast & undemanding frequency estimation (p & not-q) –If you open fridge, light goes on Geiger and Oberauer (2007, M&C) “Counterexamples play no role!”
12
Classic memory procedure Validation of Counterexample framework Probe semantic activation after reasoning –Access memory trace: FACILITATION –Inhibit memory trace: INTERFERENCE … “RAIN” !!! … … “EXERCISE” …
13
Procedure Conditional reasoning task –Valid and Invalid problems MP/MT, & AC/DA Lexical decision task –Is letter string word or not? –E.g., ‘ball’, ‘gorq’, … 50 % non-words 50 % words -50 % TARGETS: possible counterexamples -50 % UNRELATED If you water plants, they grow well. The plants grow well. Conclusion: You watered the plants 1.Conclusion is logically valid 2.Conclusion is logically invalid RAIN
14
Design experiment Conditional reasoning: Valid problems (MP/MT) Lexical Decision –Target: blank, broken, safety, jammed, unloaded… –Unrelated: onion, author, monkey, … If the trigger is pulled, the gun fires The trigger is pulled The gun fires
15
Design experiment Conditional reasoning: Invalid problems (AC/DA) Lexical Decision –Target: rain, shower, hosed, bath, splashed, … –Unrelated: writer, zebra, fare, gangster, … If Mary jumps in the pool, she gets wet Mary gets wet Mary jumped in the pool
16
Design experiment Conditional reasoning task –8 problems 4 valid (MP/MT) & 4 invalid (AC/DA) –Logical instructions Lexical decision task –20 trials 10 non-words 5 targets 5 unrelated –Target selection Based on previous generation studies –Strongly associated disabler/alternatives or associated words –Single words –Length/word frequency matched N –20 particpants
17
Design: Material List of Conditionals for the Reasoning Task Valid problems : 1.If the trigger is pulled, then the gun fires. (MP) 2.If the correct switch is flipped, then the porch light goes on. (MT) 3.If the ignition key is turned, then the car starts. (MP) 4.If the match is struck, then it lights. (MT) Invalid problems: 5. If Bart's food goes down the wrong way, the he has to cough. (DA) 6. If Mary jumps in the swimming pool, the she gets wet. (AC) 7. If the apples are ripe, then they fall from the tree. (DA) 8. If the water is poured on the campfire, then the fire goes out. (AC)
18
Prediction 1. Facilitation … Less time to recognize target words after invalid problems … 2. Inhibition … More time to recognize target words after valid problems …
19
Results Reasoning performance –Valid problems: 85% correct (SD =.22) –Invalid problems: 69% correct (SD =.26) Lexical decision data
20
Results: Lexical decision
21
Conclusion Direct memory evidence for counterexamples Retrieval and inhibition framework validated –Conditional reasoning People search for alternative causes People inhibit activated disabling conditions
22
Future work Correct vs. incorrect responses –Present study High % correct, small intra/inter individual variation –New study Large N, No logical instructions –Prediction Valid: correct slower lexical decision time Invalid: correct faster lexical decision time Children –Less inhibitory capacities, less efficient search –… Impact on semantic activations ? …
23
Thanks Angela Badulescu (York University, Toronto) Vinod Goel (York University, Toronto)
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.