Week 8: Model Answer
Tsay (2013) sets out to test the influence of vision on auditory judgements, using the domain of classical music. Are people, including experts, influenced in their judgement of sounds by the appearance of the performer?
First Tsay tested people's beliefs about how they might best judge who won a live music competition. It is not clear if the full protocol of the subsequent experiments was explained to participants in this first experiment.
Next, in a series of experiments, Tsay tested if professional musicians and non-expert participants could pick competition winners, based on 6-second clips of the 3 finalists. Conditions were either sound-only clips vs video-only clips (expts 2 and 4) or sound-only vs video-only vs sound and video (expts 3 and 5).
Finally, two further experiments aimed to investigate the mechanism by which the effect found in experiments 2-5 operated. Experiment 6 used altered stimuli, so that detail was removed but motion information was preserved in the video clips. Experiment 7 altered the dependent variable, asking participants to report the most confident, creative, involved, motivated, passionate, and unique performer.
A number of supplementary experiments were also performed, in order to check possible confounds. For space reasons I cannot cover them here.
Experiment 1: The majority (~60%) participants, all non-experts, chose sound recordings to base their judgements of who won the competition
Experiments 2-5: Experts and non-experts were able to select competitions winners at above chance rates when they had access to video-only information (both sound-only and sound-and-video conditions showed reduced performance with respect to the video-only condition, with sound-only conditions below chance and sound and video conditions being not significantly different from chance). Notably expert performance was not better than non-expert performance.
Experiment 6: Found that participants could select the winner from video clips that had restricted detail but preserved motion information
Experiment 7: Found that participants ratings of performer 'passion' predicted who won.
CONCLUSION, IMPLICATIONS AND CRITICISMS
The strengths of this study is that it confirms, using repeated experiments, that the phenomenon of visual bias is real. Cash incentives (experiment 1), clips of real music competition entrants (experiments 2-7) and the use of professional musician participants (experiments 4 and 5) raise the motivation of participants and the ecological validity of their judgements. The research addresses a real world problem - for example, the proportion of women in orchestras increased from 5% to 25% after blind auditions were introduced, testifying to the reality of how musicians look unfairly affecting their professional chances (Goldin & Rouse, 2000). It is also admirable that the research goes beyond establishing the existence of the phenomenon and attempts to uncover the psychological mechanism of action (experiments 6 and 7). In general the research also addresses the important topic of cross-modal interactions (Driver and Spence, 1998).
However, the major flaw with this research is the interpretation offered, which is profoundly misleading. The task used allows the experiments to demonstrate the existence of a visual bias on the judgements of musicians - it does not support the claim - which is made in the paper - that vision is more important than audition for judgements of musicians. To understand why this over-general, counter-intuitive, conclusion is also plain wrong consider the difference between judging absolute and judging relative quality. Sound is clearly important for judging absolute quality - no matter how good I look I cannot win a music competition playing mere "noise" (p.14583). However, if two pieces of music are closely matched, it is possible - as happens in this case - for looks to influence relative judgements. In terms of absolutely quality, and information on this quality, the stimuli in this experiment suffer a floor effect - participants are unable to discriminate quality . This is suggested by the brevity of the clips (6 seconds) and their nature (finalists in a professional music competition- the best of the best presumably). It is confirmed by the comparison of experiments 2/3 and 4/5 - experts are no more able to do this task that non-experts, suggesting that musical skill is playing no role. What participants are able to do is select winners, only using visual information. This is because - in the absences of any detectable differences in sound quality - vision biases judgements. Presumably participants are able to select winners when looks allow them to align their biased judgements with the same biases the judges in the competition possessed. Experiments 6 and 7 suggest than in these circumstances, how the performer looks in terms of movement and passion plays a biasing role.
My reinterpretation does not contradict the reality of the experimental results. If we were to run the same experiments I believe we would find the same thing. It does not contradict the importance of the result. If you are in the final of a competition it obviously matters how you look. What it does mean is that the conclusion that vision dominates audition cannot be generalised to other judgement of music. In reality - as we always believed - how music sounds matters.
Driver, J., & Spence, C. (1998). Attention and the crossmodal construction of space. Trends in cognitive sciences, 2(7), 254-262.
Goldin, C. & Rouse, C. (2000). Orchestrating Impartiality: The Impact Of 'Blind' Auditions On Female Musicians. American Economic Review, 2000, 90, 715-741.
Tsay, C. J. (2013). Sight over sound in the judgment of music performance. Proceedings of the National Academy of Sciences, 110(36), 14580-14585.