Download presentation
Presentation is loading. Please wait.
Published byBarbara Chambers Modified over 8 years ago
1
Fluency, the Feeling of Rightness, and Analytic Thinking Valerie Thompson Gordon Pennycook Jonathan Evans Jamie Prowse Turner
2
Metacognitive Reasoning Theory Dual Process Theories Automatic Type 1 processes produce a default answer that may or may not be further analysed by deliberate Type 2 processes What processes monitor Type 1 outputs? When is the default accepted or rejected? Feeling of Rightness (FOR) Accompanies Type 1 output Feeling of certainty about that output Akin to other metacognitive experiences, such as Feeling of Knowing, Judgement of Learning, etc.
3
FOR and Type 2 engagement FOR should predict P(Type 2) processing: Weak FOR → high P(Type 2) Strong FOR → low P(Type 2) → compelling illusions (Wason, Linda) FOR, like other metacognitive experiences should depend on fluency of retrieval Fluency: Ease with which answers come to mind More fluent → Strong FOR Less fluent → Lower FOR
4
Evidence for Metacognitive Reasoning Theory (Thompson, Pennycook, & Prowse Turner, under review) Conditional inference task if p, then q; p. q? N = 64, familiar materials, varied belief by validity, logic instructions, yes/no response Conditional syllogisms If p, then q; if q, then r; p. r? N = 48, non-sense middle terms, varied belief by validity, logic instructions, yes/ no response Base rate task De Neys & Glumicic (2008): personality descriptions + baserate N = 128, congruent, incongruent, neutral problems, no normative instructions, estimated probability
5
First/ intuitive answer FOR Problem Final answer FJC Final answer FJC Experimental Group Free Control Problem Two Response Paradigm Rethinking Time Answer Change
6
Evidence Summary FOR predicts Type 2 engagement: Rethinking time FOR ↑ → ↓ Rethinking time Probability of changing initial answer FOR ↑ → ↓ Answer change (probability and degree) FOR predicted by fluency to generate initial answer time to generate Answer 1 ↓ →↑ FOR Therefore, if fluency → FOR and FOR → Type 2 processing, then: Ho: Manipulating fluency should affect Type 2 processing 1. Fluency of processing 2. Fluency of production
7
Processing Fluency and Type 2 Engagement Alter et al. (2007) manipulated fluency of processing Experiment 1: CRT (Fredericks, 2005) clear (M = 1.9) vs degraded text (M = 2.5) Experiment 3: Representativeness and base rates (Tom W) puffed cheeks (r =.43) vs furrowed brow (r =.74) separate group answered trivia; puffed more confident Experiment 4: Syllogisms clear (M = 43%) vs degraded text (M = 64%) Separate group estimated difficulty without solving; clear less difficult Concluded fluency → Type 2 processing
8
Processing Fluency and Type 2 Engagement Alter et al. (2007) manipulated fluency of processing Experiment 1: CRT (Fredericks, 2005) clear (M = 1.9) vs degraded text (M = 2.5) Experiment 3: Representativeness and base rates (Tom W) puffed cheeks (r =.43) vs furrowed brow (r =.74) separate group answered trivia; puffed more confident Experiment 4: Syllogisms clear (M = 43%) vs degraded text (M = 64%) Separate group estimated difficulty without solving; clear less difficult Concluded fluency → Type 2 processing Mediated by FOR?
9
Experiment 1 (and 2) with Gordon Pennycook and Jamie Prowse Turner 16 conditionals x 4 inferences = 64 items Two responses Two groups (N = 48): Clear text Difficult text Experiment 2 (N = 64) increased difficulty, included CRT Pooled data Measures of Type 2 engagement should be lower for the fluent than non-fluent condition DVFluentNon Fluent t FOR (/7)5.435.49< 1 RT-1 (sec) 7.247.52< 1 Correct (%) 55 < 1 Rethink (sec) 6.435.62< 1 Change (%) 12101.20 Accept-1 (%) 71642.21* Accept-2 (%) 66621.17
10
Experiment 1 (and 2) with Gordon Pennycook and Jamie Prowse Turner 16 conditionals x 4 inferences = 64 items Two responses Two groups (N = 48): Clear text Difficult text Experiment 2 (N = 64) increased difficulty, included CRT Pooled data Measures of Type 2 engagement should be lower for the fluent than non-fluent condition DVFluentNon Fluent t FOR (/7)5.435.49< 1 RT-1 (sec) 7.247.52< 1 Correct (%) 55 < 1 Rethink (sec) 6.435.62< 1 Change (%) 12101.20 Accept-1 (%) 71642.21* Accept-2 (%) 66621.17 CRT.78.96< 1
11
Experiment 1 (and 2) with Gordon Pennycook and Jamie Prowse Turner 16 conditionals x 4 inferences = 64 items Two responses Two groups (N = 48): Clear text Difficult text Experiment 2 (N = 64) increased difficulty, included CRT Pooled data Measures of Type 2 engagement should be lower for the fluent than non-fluent condition DVFluentNon Fluent t FOR (/7)5.435.49< 1 RT-1 (sec) 7.247.52< 1 Correct (%) 55 < 1 Rethink (sec) 6.435.62< 1 Change (%) 12101.20 Accept-1 (%) 71642.21* Accept-2 (%) 66621.17 CRT.78 1.9.96 2.5 < 1
12
Experiment 1 (and 2) with Gordon Pennycook and Jamie Prowse Turner * * * * t (107) > 10.9, p <.001
13
Fluency of Responding vs Fluency of Processing Ho: FOR is determined by the fluency with which initial answer produced Manipulating fluency of processing should affect FOR only if it affects fluency of production E1 and E2: text manipulation had no effect on fluency of production Experiment 3: Manipulate fluency of production
14
Experiment 3 with Jonathan Evans Matching bias in Wason task Full matching paradigm, arbitrary content e.g., if the letter is not V, then the number is 8 Implicit negations Cards presented one at a time Do you need to turn this card over? 32 trials (4 rules x 4 cards x 2 repetitions) Two-response paradigm 24 U. of Saskatchewan students
15
Relative to non- matching, matching cards should be: “selected” more often responded to faster (fluency) promote stronger FOR’s engender less T2 processes: rethinking time probability of change number correct? Experiment 3 with Jonathan Evans DVMatchNon- Match t “Select”-1 (%) 71504.11** “Select”-2 (%) 69552.81* RT-1 (sec)5.706.825.26** FOR (/7)5.264.876.49** Rethink (sec) 6.016.992.35* Change (%) 9.818.33.90** Correct (%) 61 < 1
16
Experiment 3 with Jonathan Evans ** * * t (21) > 4.20, p <.001
17
Conclusions and Future Directions FOR predicts Type 2 engagement Rethinking time, probability of changing answers Fluency of responding affects FOR Manipulations that decrease speed of responding: decrease FOR’s increase Type 2 engagement Fluency of processing per se does not affect FOR’s Hard vs easy to read text did not affect FOR’s or Type 2 judgments But may affect Type 2 processing (Alter et al, 2007) Differences in participant population, method, number of problems Next: Fluent and disfluent versions of the CRT Syllogisms: accessibility of conclusion Matching in the truth table task
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.