Orientierungseffekte Inversionseffekt Orientierungseffekte Thatcher Illusion Rotierte Gesichter sind schwieriger zu erkennen als andere mono-orientierte Objekte (Yin, 1969). Der groteske Gesichtsausdruck, welcher durch das Rotieren von Augen und Mund entsteht, ist bei einem auf dem Kopf stehenden Gesicht kaum mehr zu erkennen (Thompson, 1980). WS 01/02 A. Schwaninger 3.1 WS 01/02 A. Schwaninger 3.2
Orientierungseffekte Composite Face Illusion Hypothesen zum Inversionseffekt Konfigural-Komponenten-Hypothese Übereinandergelegte Teile verschiedener Gesichter wirken bei aufrechter Präsentation wie ein neues Gesicht und die zwei Identitäten sind schwieriger zu erkennen. Durch seitliches Versetzen und Rotation um 180 verschwindet der Effekt (Aus Stevenage, 1995; nach Young et al., 1987). Component Processing eher orientierungsinvariant Configural Processing sehr orientierungssensitiv WS 01/02 A. Schwaninger 3.3 (E.g. Bartlett & Searcy, 1993; Carey & Diamond, 1977; Diamond & Carey, 1986; Leder & Bruce, 1998; Searcy & Bartlett, 1996; Sergent, 1984) WS 01/02 A. Schwaninger 3.4
Hypothesen zum Inversionseffekt Hypothese holistischer Verarbeitung Hypothesen zum Inversionseffekt Mentale Rotationshypothese Faces Objects Words Holistic Part-based (Nach Gazzaniga et al., 1998) "Face perception is holistic and the perception of holistically represented complex patterns is orientation sensitive." (Farah et al., 1995, p.633). Definition "holistic": Keine explizite Repräsentation der Teile (Bsp. Bitmap). Rock gibt eine Erklärung für die Thatcher-Illusion 7 Jahre vor ihrer Entdeckung: "In this [inverted] situation, there is a whole set of component figures and figural relationships to be corrected, and it is not possible to succeed in visualizing simultaneously how each of these would look were it to be egocentrically upright (Rock, 1973, p.60f.)". WS 01/02 A. Schwaninger 3.5 WS 01/02 A. Schwaninger 3.6
Hypothesen zum Inversionseffekt Multidimensional Space Framework Experiment 1 Component Changes Experiment 2 Configural Changes "Presenting a face upside-down is an example of one experimental manipulation that would make the encoding conditions difficult, leading to a large error associated with the location in the multidimensional space derived from a stimulus face" (Valentine, 1991, p.172). Eyes and Mouth Replaced Inter-Eye Distance and Eye- Mouth Distance Increased WS 01/02 A. Schwaninger 3.7 WS 01/02 A. Schwaninger 3.8
Voraussagen der 5 Hypothesen Sequentieller Same-Different Task Nullstimulus 3 sec Pause 1 sec Teststimulus: same or different? Button Press Pause 1 sec 84 trials per experiment: 6 (faces) * 7 (angles) * 2 (same/different) WS 01/02 A. Schwaninger 3.9 WS 01/02 A. Schwaninger 3.10
Errors for the Processing of Component and Configural Alterations 50% Response Times for the Processing of Component and Configural Alterations 1500 Errors [%] 40% 30% 20% 10% RT [ms] 1300 1100 900 0% 0 30 60 90 120 150 180 Angle of Rotation 700 0 30 60 90 120 150 180 Angle of Rotation Two factor ANOVA on error scores Main effect of angle [p < 0.001] Main effect of experiment type [p < 0.001] Significant interaction [p < 0.001] Separate polynomial contrast analyses Component changes: linear component [p < 0.05] explaining 14.5 % of the variance Configural changes: quadratic component [p<0.001] explaining 65 % of the variance WS 01/02 A. Schwaninger 3.11 Two factor ANOVA on correct RTs Main effect of angle [p < 0.001] No main effect of experiment type [p = 0.666] No interaction [p = 0.176] Separate polynomial contrast analyses Component changes: linear component [p < 0.001] explaining 60.4 % of the variance Configural changes: linear component [p < 0.001] explaining 61.2 % of the variance WS 01/02 A. Schwaninger 3.12
Group 1 (n = 32) Group 2 (n = 32) 1st Exp. component changes 2nd Exp. configural changes TRANSFER EFFECTS 1st Exp. configural changes 2nd Exp. component changes "Why is face recognition so orientation sensitive? Transfer effects can reveal whether two abilities are based on independent processing systems or whether they share common mechanisms. Transfer type Trial type Transfer effect error scores Transfer effect reaction times configural different no no components same yes no components different yes yes configural) same yes no Rotated faces can not be processed as unparsed wholes ( holistically ). Rotated faces have to be processed by mentally rotating facial parts. (4) Several transfer effects have been revealed [all p values < 0.05] The processing of component and configural information shares common mechanisms, including mental rotation. WS 01/02 A. Schwaninger 3.13 WS 01/02 A. Schwaninger 3.14
Mögliche Integration der Hypothesen Die Frage nach der Spezifität Rotationseffekte spezifisch? (Aus Rock, 1973, 1974) WS 01/02 A. Schwaninger 3.15 WS 01/02 A. Schwaninger 3.16