Introduction: Non-rigid and rigid motion have been shown to facilitate face recognition (e.g. Thornton & Kourtzi, 2002; Lander & Bruce, 2000; Lander et al, 1999; Watson et al, submitted, Pilz et al, in preparation). Another familiar kind of motion occurs when a person approaches you. Does this kind of looming motion have any effect on identity decisions? Karin Pilz, Ian M. Thornton, Quoc C. Vuong & Heinrich H. Bülthoff Max Planck Institute for Biological Cybernetics, Tübingen, Germany karin.pilz@tuebingen.mpg.de Recognising Faces in Motion - Old/new recognition tasks may not be appropriate for investigating identity processing with dynamic stimuli - There is a significant reaction time advantage for matching looming figures as compared to static snapshots. - Part of this advantage might be due to enhanced arousal levels provided by the motion of the stimulus. - However, as we cannot find a significant difference between dynamic and static conditions in the control experiment, the looming of the person must also play a role. ﾆ Looming not only affects working memory performance but also facilitates identity perception over time as subjects are more efficient in searching for dynamically learned faces. Conclusions Lander K, Bruce V (2003). The role of motion in learning new faces. Visual Cognition, 10, 897-921. Lander K, Christie F, & Bruce V (1999). The role of movement in the recognition of famous faces. Memory and Cognition, 27, 974-985. Pilz K, Thornton I M & Bülthoff H H (2004). The dynamic face advantage. In preparation Thornton I M & Kourtzi Z (2002). A matching advantage for dynamic faces. Perception, 31, 113-132. Watson T L, Hill H C, Johnston A & Troje N (2004). Motion as a cue for viewpoint invariance. Submitted References * Looming Person Attention Control (looming dots)  Looming Person (within subject design): We used a sequential matching paradigm to investigate whether the looming motion of a person facilitates matching. In half of the trials the prime face was a looming figure, in the other half of trials a static snapshot. The target face was always a coloured snapshot of a person’s head out of frontal view or 45°on either side of the face. Subjects (N=12) were told to respond as quickly and accurately as possible on whether the prime and the target were of the same identity or not. Control (looming background dots) ﾆ In the control experiment the prime was always a static snapshot presented on a random dot background. In one condition the random dot pattern was moving towards the observer, whereas it remained static in the other condition. Subjects (N=12 ) again had to match the identity of prime and target face. Sequential Matching Reaction time difference (ms) between static and dynamic prime faces for same and different trials for all subjects. T-testing revealed a significant difference between the two conditions (p = 0.05). Difference static - dynamic trials (Looming Person) 0 5 10 15 20 25 30 35 40 45 Same Trials Different Trials RT (ms) Reaction time difference (ms) between static and dynamic prime faces for same and different trials for all subjects. T-testing revealed NO significant difference between the two conditions. Difference static - dynamic trials (Control) 400 450 500 550 600 650 700 Same Trials Different Trials RT(ms) Reaction times (ms) for same and different trials for all subjects for both dynamic (blue) and static (green) prime faces. Subjects are significantly faster (36 ms) for dynamic primes in the ‘same’ condition (p < 0.005). Looming Person 400 450 500 550 600 650 700 Same Trials Different Trials RT (ms) Reaction times (ms) for same and different trials for all subjects for both dynamic (blue) and static (green) prime faces. Subjects are significantly faster (20 ms) for dynamic primes in the ‘different’ condition (p < 0.05). Control (looming dots) 0 5 10 15 20 25 30 35 40 45 Same Trials Different Trials RT (ms) * * Recognition Methods Results and Conclusions Learning Phase (between subject design): Subjects were familiarised with 12 individuals, either shown as static images or whole body figures approaching them. They filled out a questionnaire concerning personality and characteristic facial features for each person. Test Phase: After a distraction task, subjects performed a 2AFC recognition task. Two frontal view faces were presented on the screen and subjects had to identify as quickly and accurately as possible, the person they had been familiarised with during learning. - Previous experiments using similar tasks to investigate the impact of non-rigid motion on face recognition have also revealed contradictory results. It is possible that such explicit recognition paradigms may be inappropriate for studying dynamic aspects of object representation. Below we explore the use of other types of tasks. - There was no significant difference between conditions. While pilot data showed a significant accuracy advantage for dynamic trials, this was not replicated in a full design with accuracy constant around 65 %. 0 10 20 30 40 50 60 70 80 Static Moving Type of familiarisation Percent correct Visual Search Test Phase Familiarisation Phase Target 1 Target 2 Blank 1000 ms 2000 ms 1000 ms 100 alternations Test Phase: After a short break, subjects performed a visual search task. Two to six static faces were presented on the screen and subjects had to decide as quickly and accurately as possible, whether one of the familiarised faces was present. - Subjects were more efficient (left) in detecting the dynamically learned stimulus (blue). - Search slopes (right) for dynamically learned faces were significantly shallower (p < 0.05) than search slopes for the statically learned faces. -Accuracy data also showed a trend for subjects to be better at searching for the dynamically learned stimulus (dynamic = 90.63%, static = 83.33%, p < 0.7) Learning Phase: Subjects (N=7) were familiarised with one static image of a person and one whole body figure approaching them. Methods Results 4000 ms Test Phase Learning Phase Looming or static person as a prime (800 ms) 300 ms blank Target face until response Prime face with moving or static background (800 ms) 300 ms blank Target face until response * 0 50 100 150 200 250 Moving Static Slope (ms/ item) 600 800 1000 1200 1400 1600 1800 2000 2200 2 4 6 Set Size RT (ms)