Nonprobative photos inflate the truthiness and falsiness of claims

Newman, E. J. (2013). Nonprobative Photos Inflate the Truthiness and Falsiness of Claims (Version 1). Open Access Te Herenga Waka-Victoria University of Wellington. https://doi.org/10.26686/wgtn.17003764.v1

Abstract de l'autrice

When people evaluate claims they often rely on what comedian Stephen Colbert calls truthiness, judging claims using subjective feelings of truth, rather than drawing on facts. Over seven experiments I examined how nonprobative photos can manufacture truthiness in just a few seconds. I found that a quick exposure to a photo that relates to, but does not provide any probative evidence about the accuracy of claims can systematically bias people to conclude claims are true. In Experiments 1A and 1B, people saw familiar and unfamiliar celebrity names and, for each, quickly responded "true" or "false" to the claim "This famous person is alive" or (between subjects) "This famous person is dead." Within subjects, some names appeared with a photo of the celebrity engaged in his/her profession whereas other names appeared alone. For unfamiliar celebrity names, photos increased the likelihood that subjects judged the claim to be true. Moreover, the same photos inflated the truth of "Alive" and "Dead" claims, suggesting that photos did not produce an "alive bias," but a "truth bias."

Experiment 2 showed that photos and verbal information similarly inflated truthiness, suggesting that the effect is not peculiar to photographs per se. Experiment 3 demonstrated that nonprobative photos can also enhance the truthiness of general knowledge claims (Giraffes are the only mammals that cannot jump). In Experiments 4-6 I examined boundary conditions for truthiness. I found that the semantic relationship between the photo and claim mattered. Experiment 4 showed that in a within-subject design, related photos produced truthiness, but unrelated photos acted just like the no photo condition. But unrelated photos were not always benign, Experiment 5 showed that their effects depended on experimental context. In a mixed design, related photos produced truthiness and unrelated photos produced falsiness.

Although the effect of related photos was robust across materials and variation in experimental context, when I used a fully between-subjects design in Experiment 6, the effect of photos (related and unrelated) was eliminated. These effects add to a growing literature on how nonprobative information can influence people’s decisions and suggest that nonprobative photographs do more than simply decorate, they can rapidly manufacture feelings of truth. As with many effects in the cognitive psychology literature, the photo-truthiness effect depends on the way in which people process and interpret photos when evaluating the truth of claims.

Chapter 1

In this thesis I use the term “truthiness” to refer to a category of phenomena in which the addition of nonprobative information causes people to shift rapidly towards believing a claim (see also the illusory truth effect; Begg, Anas, & Farinacci, 1992).

Overview

If photos help boost cognitive availability of related thoughts and images, or help people trawl for and selectively interpret information in line with their hypothesis about unfamiliar celebrities, then photos should increase the truthiness of claims about those celebrities, regardless of whether the claim is that the celebrity is alive or dead (cf. Unkelbach, 2007).

Chapter 2 - Experiment 1A & 1B

Experiment 1A and 1B

Design

• VI1: photograph: yes, no
• VI2: familiarity: familiar, unfamiliar (i.e. est-ce que la personne est "connue" ou non)
• VI3: claim: alive, dead
• VD: C score (déviation à dire "true")

Résultant en un 2x2x2 mixed design, avec photograph et familiarity within subjects et le claim between subjects.

Résultats

• Elle a réalisé une mixed repeated measures (photograph) x 2 (familiarity) x 2 (claim) ANOVA
• Effet de la photo:
  • Experiment 1A, F(1, 90) = 4.87, p = .03, η_p² = .05;
  • Experiment 1B, F(1, 46) = 10.53, p < .01, η_p² = .19)

Bias for claims about familiar and unfamiliar names (in Experiments 1a-1b) presented with or without a photograph and collapsed across dead/alive factor. Negative value of c is a bias to say true. In Experiments 1a-2, photos (or words) affected bias for unfamiliar names. Error bars show 95% within-subject confidence intervals for the photo/no-photo effect at each level of familiarity/difficulty (see Masson & Loftus, 2003)

Discussion

Newman remarque que les gens ont tendance à trouver vraies les "Alive" claims plus souvent que les "Dead" claims.

A ce moment, Newman se demande si c'est un biais de truthiness ou d'aliveness (i.e. le fait qu'on voit la photo fait penser que la personne est en vie).

Photos might promote truthiness because they provide a rich semantic context that facilitates the generation of thoughts and images relating to the claim. But these results also fit with the idea that feelings of truthiness arose because photos are inherently credible or that over time people have come to associate photos with the conclusion that something is real and true (cf. Unkelbach, 2007).

Chapter 3 - Experiment 2 and 3

Experiment 2

Pour l'expérience 2, elle se demande si des types d'informations non probantes autres que les photos vont avoir un impact sur la truthiness. Elle a donc cherché à comparer les types de formats (photo et verbal).

Design

Variables:

• VI 1: nonprobative information: yes, no
• VI 2: format of nonprobative information: photo, verbal
• VI 3: claim: alive, dead
• VD : C score

Elle a abouti à un design mixte 2 (nonprobative information) x 2 (format of nonprobative information) x 2 (claim) avec le format et le claim comme élément intrasujet. Elle a aussi gardé uniquement les unfamiliar names.

Résultats

Elle a fait une mixed repeated measures 2 (nonprobative information: yes, no) x 2 (format of nonprobative information: photo, verbal) x 2 (claim: alive, dead) ANOVA. Elle a trouvé un effet de la présence de nonprobative information: F(1, 50) = 10.27, p < .01, η_p² = .17.

Bias for claims about unfamiliar names (Experiment 2) presented with or without a photograph or verbal information, collapsed across dead/alive factor; easy and difficult trivia claims (in Experiment 3) presented with or without a photograph . Negative value of c is a bias to say true. In Experiment 2, photos (or words) affected bias for unfamiliar names; in Experiment 3, photos affected bias for difficult trivia statements. Error bars show 95% within-subject confidence intervals for the photo/no-photo effect at each level of familiarity/difficulty (see Masson & Loftus, 2003).

Discussion

Elle a donc trouvé un effet de la nonprobative information, que ce soit du texte ou de l'image. Ce n'est donc pas directement lié à la crédibilité de l'image.

Experiment 3

Elle regarde l'impact de l'image sur des "trivia claims". Elle utilise de phrases produites pour les articles de Unkelbach (2007) et Nelson & Narens (1980).

Design

Variables:

• VI1: photograph: yes, no
• VI2: difficulty: easy, hard
• VD: C score

Ce qui en fait un design (photograph: yes, no) x 2 (difficulty: easy, hard) within-subjects design

Résultats

Elle a fait une repeated measures 2 (photograph: yes, no) x 2 (difficulty: easy, hard) ANOVA Elle a trouvé un effet de la photo : F(1, 69) = 6.65, p = .01, η_p² = .09

Summary of Experiments 1-3

Photos did not produce an "alive bias," but a "truth bias." Moreover, the truthiness effect generalized beyond Dead or Alive judgements: Nonprobative photos enhanced the subjective truth of general knowledge claims, too.

Effect of photos on subjective truth is driven not simply by a perception that photos are inherently trustworthy. We speculate that nonprobative photos and verbal information help people generate pseudoevidence (cf. Kelly & Nace, 1994).

our findings suggest that even without repeated exposures or instructions to imagine, the mere presence of nonprobative information such as photos might rapidly inflate the perceived truth of many types of true and false claims

Under what conditions do photos produce truthiness?

There are reasons to predict that relative to related photos, unrelated photos should cause disfluency. That is, unrelated photos should make it more difficult for people to bring related thoughts and images to mind. For instance, in one study, when people saw a target word that was semantically unrelated to a preceding sentence (“The librarian reached for the top shelf and pulled down a book” “napkin”), they rated that target word as less pleasant than when they saw a target that was semantically related (“read”; Lee & Labroo, 2004)

(...) The semantically unrelated words were relatively disfluent compared to their semantically related counterparts.

(...) Parallel effects occur with perceptual manipulations; degraded images, difficult fonts, and low contrast colors can lead people to invest more cognitive effort to make sense of stimuli and people tend to be evaluate these stimuli more negatively (Diemand-Yauman, Oppenheimer, & Vaughan, 2011; Petrova & Cialdini, 2005; Reber & Schwarz 1999; see Yue, Castel, & Bjork, in press).

(...) That is, nonprobative, unrelated information should cause people to disbelieve a claim

(...)

But there are also reasons to predict that unrelated photos would not cause falsiness

(...) adding irrelevant information produces the same effect as giving people no information at all (Gill, Swann, & Silvera, 1998). Thus, a trawling mechanism would predict that unrelated photos would have little if any effect on people’s decisions.

Chapter 4

Experiment 4

Résultats

Bias for difficult trivia claims presented with or without a photograph that was semantically related or unrelated to the trivia claims. A negative value of c is a bias to say true. Error bars show 95% within-subject confidence intervals for the photo factor for Experiment 4 (see Masson & Loftus, 2003)

Experiment 5

Résultats

Bias for difficult trivia claims presented with or without a photograph that was semantically related or unrelated to the trivia claims. A negative value of c is a bias to say true. Error bars show 95% within-subject confidence intervals for the photo effect for related and unrelated conditions in Experiment 5 (see Masson & Loftus, 2003)

Experiment 6

Résultats

Bias for difficult trivia claims presented with or without a photograph that was semantically related or unrelated to the trivia claims. A negative value of c is a bias to say true. Error bars show 95% confidence intervals for each cell mean in Experiment 6 (see Masson & Loftus, 2003)