Download presentation
Presentation is loading. Please wait.
Published byJudith Reynolds Modified over 8 years ago
1
Default logic and effortful beliefs Simon Handley Steve Newstead
2
Dual processes Stanovich (1999) Evans (2006), Evans & Over (1996) Sloman (1996) etc. Pre-conscious processes cue contextualised representations. These may lead directly to default responses based upon beliefs, heuristics, pragmatics etc. The analytic system may intervene given sufficient resources or time
3
Is belief based reasoning really System 1? Children’s reasoning –Morsanyi & Handley (2008) –Markovits & Barrouillet Adult reasoning –de Neys et al (2005) –Mckinnon & Moscovitch (2007) –Apperly et al. (2006) –Evans, Handley & Bacon (2008) –Newstead, Handley et al. (2004)
4
Reasoning and instruction Typically participants are asked to assume the premises etc. Participants must understand and maintain instructional set. In this experimental series we introduced a second instructional set – belief instructions –e.g - respond on the basis of what you believe to be true in the world.
5
Reasoning and belief-logic conflict Conflict Problem If a child is crying then the child is happy The child is crying Is the child happy? Logic – Yes Belief - No No Conflict problem If a child is crying then the child is sad The child is crying Is the child happy? Logic – No Belief - No
6
Reasoning and beliefs – some dual process predictions Default – Interventionist predictions (beliefs are primary) Reasoning based upon beliefs should be quicker than reasoning based upon logic Conflict problems will be more difficult/ take longer to process than non-conflict problems Beliefs will interfere with logical processing, but logic should not interfere with belief based reasoning. More errors expected under logical instructions
7
Belief Instructions “ …this means that you must answer according to your knowledge of what you know to be true in the world; For example:If you finish a drink then the glass will be full Suppose that you finish your drink Will it be full? Based upon our knowledge of the world we know that if a drink is finished then the glass would be empty, so the correct answer is No”
8
Logic Instructions “……this means you must assume that each premise is true (even if in reality it is not true) and respond with the answer that logically follows from the statements presented. For example:If you finish a drink then the glass will be full Suppose that you finish your drink Will it be empty? The logical answer to this problem is No, because you must assume that if a drink is finished the glass is full, therefore the glass will not be empty when the drink is finished.
9
Design 30 participants 80 conditional arguments (MP) Two blocks of problems –Group 1: Logic followed by Belief instructions –Group 2: Belief followed by Logic instructions Identical problems presented in each block –10 CTF – Logic answer Y, Belief answer N –10 CTF – Logic answer N, Belief answer Y –10 Factual – Logic answer Y, Belief answer Y –10 Factual – Logic answer N, Belief answer N Two problem sets
10
Latency Results
11
Accuracy data
12
Summary Overall belief judgments take less time than logic judgments, BUT some indication that conflict has a greater effect on latencies for belief judgments. Belief judgments lead to more error, but only on belief/logic conflict problems Experiment 2 –Replicate findings of Experiment 1 – in terms of error and latency effects –Adopted a trial by trial instructional cueing method
13
Experiment 2 Different set of problems with identical structure Mixed trial type design with randomly ordered but equal number of logic and belief instruction trails Pre training session on different instructional sets (same participants as Experiment 1) Sequence of events –500ms – fixation point –Prime (L or B) in centre of screen for 1000ms –Problem presentation –40 trials
14
Latency analysis
15
Accuracy data
16
Summary Experiment 2 Judgments under belief instruction quicker than under logical instructions Increased error rates under belief instructions Conflict between belief and logic has a greater impact on belief judgments than logic judgments – for both accuracy and latency
17
Experiment 3 Replication of Experiment 2 Instructional cue presented 250ms or 500ms prior to problem presentation.
18
Latency analysis
19
Accuracy data
20
Experiment 4 Is the latency difference in Experiments 1 and 2 a result of strategic processing? If you finish a drink then the glass will be full Suppose that you finish your drink Will it be full? Participants only need to process the categorical premise and the conclusion to give a belief based response – the full argument must be processed for a logical response to be generated
21
Experiment 4 Instructional cue presented after problem presentation Presentation sequence –First Premise –Space Bar –Second premise, Conclusion plus response cue: Valid/Invalid Believable/unbelievable –4 versions of 48 problems –32 participants
22
Latency analysis
23
Accuracy data
24
Summary No evidence that belief based judgments take longer than logic based judgments Given, even on this version, that participants need only consult the categorical premise and conclusion if anything belief based processing may take longer As earlier studies there is a unidirectional impact of the logic of a problem on accuracy but little impact of belief content on logical judgments.
25
Experiment 5 Two problem contents: Original Nonsense middle terms If a ball is thrown in the air then it stays in the airAll Zabs can walk The ball is thrown in the airWhales are Zabs Therefore the ball drops to the floorTherefore whales can walk 48 problems, 2 problem sets, 32 participants. 8 practice problems with feedback. Trial sequence:First premise Space Bar Second premise, conclusion and instructional cue.
26
Latency analysis
27
Accuracy data
28
Experimental summary Belief based reasoning is more sensitive to conflicting logical structure than logical responding is sensitive to conflicting beliefs. This results in more errors under belief instructions: –Where instructional set is applied to problem blocks (Experiment 1) –Where instructional set is primed prior to problem presentation (Expts 2 and 3) –Where instructional set is primed following problem presentation (Expts 4 and 5) –This holds for both realistic and abstract premises (Experiment 5) –The effect is interactive (e.g. only holds for conflict problems) Paradoxically belief based reasoning is faster than logical reasoning –This holds when problems are presented in blocks –And when cue is presented prior to problem presentation –Suggests that P’s can reduce processing requirements if given advanced warning –However, Expts 4 and 5 suggest that belief based reasoning may actually take longer when position of cue prevents strategic processing
29
Conclusions Belief based reasoning on this task is not primary, heuristic and rapid Beliefs do not interfere with logical responding, but logic interferes with belief based responding The sort of logical processing involved here is primary, rapid and automatised?
30
Some final thoughts Key dual process manipulations –Speeded task –Concurrent load –Individual differences But all these manipulations also been shown to reduce contextualized responding on reasoning tasks (accessing and integrating knowledge) The impact that these manipulations have on logical reasoning may have nothing to do with dual processes but may simply reflect the capacity that individuals have to maintain an instructional set in the face of interference.
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.