Human reasoning is nonmonotonic or defeasible Additional premises can render otherwise acceptable inferences unacceptable Byrne 1989 Example of suppression effects ie acceptance ID: 562300
Download Presentation The PPT/PDF document "Suppression Effects in Conditional Reaso..." is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
Suppression Effects in Conditional ReasoningHuman reasoning is non-monotonic or defeasible: Additional premises can render otherwise acceptable inferences unacceptable (Byrne, 1989). Example of suppression effects (i.e., acceptance rates reduced with additional premises):A conditional is an "if p then q" statement. Four conditional inferences:Suppression effects in conditional inferences:Additional disablers (disabling the link of p on q) suppress acceptance of MP and MT.(Example: Exercising is a disabler that precludes weight gain)Additional alternatives (enabling q in the absence of p) suppress acceptance of AC and DA.(Example: Eating food rich in calories is an alternative way of gaining weight)Research Question: Which cognitive processes underlie suppression effects of additional disablers and additional alternatives in probabilistic conditional reasoning?
Henrik Singmann*, Karl Christoph Klauer, and Sieghard Beller
Poster presented at Psychonomic Society's 56th Annual Meeting, Chicago (2015)
Dissociating Different Suppression Effects Using the Dual-Source Model of Probabilistic Conditional Reasoning
Methods3 conditions with 178 participants in total. At least 2 weeks between the two phases. Everyday reasoning instructions.Participants provide degree of endorsement to conclusion on scale from 0% to 100%. 4 different contents (Klauer et al., 2010)Responses to all four inferences (MP, MT, AC, & DA) for each content and each phase. 16 responses per participant per phase (32 responses in total).Fitted with DSM with 16 free parameters.
ReferencesByrne, R. M. J. (1989). Suppressing valid inferences with conditionals. Cognition, 31(1), 61–83. http://doi.org/10.1016/0010-0277(89)90018-8Klauer, K. C., Beller, S., & Hütter, M. (2010). Conditional reasoning in context: A dual-source model of probabilistic inference. Journal of Experimental Psychology: Learning, Memory, and Cognition, 36(2), 298–323. http://doi.org/10.1037/a0018705Oaksford, M., Chater, N., & Larkin, J. (2000). Probabilities and polarity biases in conditional inference. Journal of Experimental Psychology: Learning, Memory, and Cognition, 26(4), 883–899. http://doi.org/10.1037/0278-7393.26.4.883
We successfully validated 2HTM and UVSD for a recognition memory task with three response options, “OLD", “UNSURE", and “NEW":
Experimental manipulations, that were expected to selectively influence memory or guessing processes, indeed affected solely the corresponding parameters of the models (i.e., memory and guessing parameters, respectively).
*singmann@gmail.com
Baseline
(
higher acceptance rates)Additional premise (lower acceptance rates)If a person drinks a lot of coke then the person will gain weight. A person drinks a lot of coke. Does it follow that the person will gain weight?If a person drinks a lot of coke then the person will gain weight. If a person does NOT exercise a lot then the person will gain weight. A person drinks a lot of coke. Does it follow that the person will gain weight?
MP (modus ponens)MT (modus tollens)AC (affirmation of consequent)DA (denial of antecedent)If p then q.p.Does q follow?If p then q.not q.Does not p follow?If p then q.q.Does p follow?If p then q.not p.Does not q follow?(valid inference)(valid inference)(invalid inference)(invalid inference)
Dual-Source Model (DSM)Extends purely Bayesian models of probabilistic conditional reasoning (Oaksford, Chater, Larkin, 2000).Assumes three independent cognitive processes/parameters (Klauer, Beller, & Hütter, 2010):Knowledge-based information (ξ): Background knowledge; beliefs about conditional probability of conclusion given minor premise; e.g., MP: P(q|p). Bayesian process , same as in Oaksford et al. model, assumes that responses to the four inferences come from one probability distribution over p and q.Form-based information (τ): Subjective probability with which an inference is seen as warranted based on the form alone; e.g., τ(MP) > τ(MT) > τ(AC) = τ(DA). Form-effect that is independent of contentMixture weight (λ): Relative weight given to form-based compared to knowledge-based component. Two phase experimental paradigm to estimate dual-source model (separated by at least two weeks):
Knowledge phase (reduced inferences)Rule phase (full conditional inferences)A person drinks a lot of coke. Does it follow that the person will gain weight?If a person drinks a lot of coke then the person will gain weight. A person drinks a lot of coke. Does it follow that the person will gain weight?Allows estimation of knowledge component (i.e., probability distribution over p and q) in absence of conditional.Allows estimation of influence of form-based component and mixture weight.
knowledge-based
form-
based
DSM
parameters
:
ξ = knowledge-based informationτ = form-based informationλ = weighting parameterC = content (1 – 4)x = inference (MP, MT, AC, & DA)
mixture
Knowledge
phase (reduced inferences)Rule phase (full conditional inferences)BaselineA person drinks a lot of coke.Does it follow that the person will gain weight?If a person drinks a lot of coke then the person will gain weight.A person drinks a lot of coke.Does it follow that the person will gain weight?AlternativesA person drinks a lot of coke.Does it follow that the person will gain weight?Please note:A person also gains weight ifthe person eats a lot,the person has metabolic problems,the person hardly exercises.If a person drinks a lot of coke then the person will gain weight.A person drinks a lot of coke.Does it follow that the person will gain weight?Please note:A person also gains weight ifthe person eats a lot,the person has metabolic problems,the person hardly exercises.DisablersA person drinks a lot of coke.Does it follow that the person will gain weight?Please note:A person only gains weight ifthe metabolism of the person permits it,the person does not exercise as a compensation,the person does not only drink diet coke.If a person drinks a lot of coke then the person will gain weight.A person drinks a lot of coke.Does it follow that the person will gain weight?Please note:A person only gains weight ifthe metabolism of the person permits it,the person does not exercise as a compensation,the person does not only drink diet coke.
Results and DiscussionReplication of suppression effects in probabilistic setting (disablers suppress MP & MT, alternatives suppress AC & DA).Observed data shows dissociation of suppression effects:Disablers mainly suppress endorsement in rule phase. Only small suppression in knowledge phase.Alternatives suppress endorsement of AC and DA equally across phases.In knowledge phase effects of alternatives larger than effect of disablers. No difference in rule phase.DSM shows dissociation in terms of underlying cognitive processes:Both type of counterexamples suppress form-parameters (τ) of "attacked" inferences:Explicit counterexamples decrease the perceived correctness of the form of the attacked inferences.Alternatives (and to a small degree disablers) suppress knowledge-based component (ξ) of "attacked" inferences:Explicit alternatives affect which background knowledge is recruited for reasoning.Disablers suppress weight (λ) given to form-based component: Explicit disablers discredit the conditional with the consequence that participants more strongly resort to their background knowledge.
*
*
*
*
*
*
*
*
*
*
Plots:
raw values in grey (background)
condition means in black
*:
p
< .05