Investigating the Fluency-Affect relationship in judgements of (in)coherent risk

Gamblin, David M. (2015) Investigating the Fluency-Affect relationship in judgements of (in)coherent risk. In: SPUDM 2015, 16-20 Aug 2015, Budapest, Hungary. (Unpublished)

Abstract

Recent accounts of emotion in intuitive judgement and decision making have separated integral affect from incidental affect (Blanchette & Richards, 2010) and demonstrated how it can be beneficial to decision making (Topolinski, 2011). However, there is disagreement concerning its mechanisms. Presenting a highly influential model, Winkielman et al (2003) suggest that fluency is hedonically marked; demonstrating that when stimuli are processed with ease, this fluency increases how much the object is liked. In contrast, Albrecht and Carbon (2004) suggest that this only occurs when stimuli are positive. They found that when rating negative stimuli, fluency increased the disliking of the object. This suggests that fluency causes more intense feelings, rather than inherently positive ones. The way in which this affect is subsequently used in judgement has also presented contrasting accounts. More traditional heuristics and biases accounts have suggested that emotion is a source of bias in judgement and decision making. For example, the Affect Heuristic may lead us to substitute difficult questions, such as, “What do I think about this?”, with the easier question, “How do I feel about it?” (Kahneman, 2011). However, recent accounts have shown that affect can be beneficial to judgement and decision making, particularly as a cue for detecting coherence (Topolinski, 2011). The current study aims, over 3 experiments, to investigate these mechanisms in order to clarify the debate; specifically it aims to answer, ‘How is integral affect evoked from the coherence of risk information?’, and ‘How is that affect is used to form a judgement of risk?’ In all 3 experiments, participants were presented with triads of risk items relating to various businesses, in a 2 (Coherence: Coherent vs Incoherent) x 2 (Risk: High vs Low) repeated measures design – with reading time of the triads as a measure of fluency, liking ratings, and judgements of risk collected as dependent variables. In Experiment 1, coherence was manipulated by varying the consistency of valence of the risk items. Coherent triads contained 3 pieces of information which all suggested either high risk or low risk, whilst incoherent triads contained a mixture of high and low risk items. Results showed that coherent triads were processed more fluently than incoherent triads, and that this fluency had an influence on liking ratings. Specifically, fluency appeared to increase the liking ratings of low risk items and increased the disliking of high risk items – a pattern that is predicted by the fluency amplification model (Albrecht & Carbon, 2014). This finding is in contrast to hedonic marking (Winkielman et al, 2003), which predicts that fluency evokes a positive response, regardless of whether the presented information related to high or low risks. In contrast, judgements of risk were not influenced by coherence, instead following a linear pattern predicted by a tallying process (Dawes, 1979). This is a more rational method, whereby the overall risk judgement was based on the number of high risk items presented in the scenario. In Experiment 2, fluency was manipulated by presenting triads of risk information that were either typical or atypical for the given businesses. Consistent with Experiment 1, results showed that coherent triads were processed with greater fluency than incoherent triads, and that judgements of risk followed a rational decision making pattern. However, participants’ liking ratings only followed the pattern from Experiment 1 when rating low risk items, whereby coherent triads were liked more than incoherent triads. When rating high risk items, no significant difference was found between coherent and incoherent triads. Finally, to test whether fluency would influence risk judgements if participants were forced to act more intuitively, a speeded version of Experiment 2 was run. Results showed that enforcing time limits caused the participants to adjust their decision making style. Critically, in contrast with the previous experiments, fluency of processing had an impact on the participants’ risk judgements in the low risk condition: with coherent triads being judged as less risky than incoherent triads. In conclusion, the study has theoretical implications by firstly demonstrating support for a fluency amplification model. Secondly, that enforcing time limits resulted in participants shifting their decision making style from a more rational process, to one which was influenced by fluency. Finally, and in contrast to Topolinski’s (2011) findings, it was fluency – and not affect – that influenced risk judgements.

Metadata

Statistics

DownloadsShow export options

Export as XMLJSONCSV

Activity Overview

6 month trend

65Downloads

6 month trend

228Hits

Additional statistics are available via IRStats2.

Archive Staff Only (login required)

Edit/View Item