Loading...
Publication
Stereotypes-as-context in visual change detection tasks
Authors
Advisor(s)
Abstract(s)
Processos de identificação e de detecção visual demonstram diferente sensibilidade às
características do contexto. No primeiro caso, estímulos congruentes com o contexto são
identificados mais rapidamente do que os incongruentes; no segundo, a incongruência é detectada
mais rapidamente. Em qualquer dos casos é clara a influência das expectativas criadas pelo
contexto.
Estudos que usam estereótipos enquanto contexto também observam identificação mais
rápida para estímulos congruentes com o contexto. Mas pouco se sabe sobre os efeitos deste tipo
de contexto em tarefas de detecção. Para testar este efeito utilizou-se uma flicker task na qual 72
participantes detectaram a mudança de um objecto em cenas naturais contendo uma mulher ou
homem. O objecto, estímulo crítico, foi congruente (e.g., homem-charuto) ou incongruente com a
pessoa (e.g., mulher-gravata). Esperava-se diferenças entre estímulos congruentes e incongruentes,
com detecções mais rápidas de estímulos incongruentes quando comparados com congruentes. Os
resultados mostram que o nível de congruência não tem impacto na velocidade de detecção. Esta
foi apenas afectada pelo tempo decorrido entre o início de cada trial e a ocorrência de mudança
(threshold): em trials em que esta ocorre aos 1200ms, a detecção é significativamente mais lenta
do que quando ocorre aos 400ms ou 800ms.
Discutem-se estes resultados à luz da literatura sobre detecção de mudança e estereótipos,
focando problemas metodológicos da tarefa e a possibilidade de que estereótipos possam
influenciar tarefas de identificação e detecção diferentemente. O efeito de threshold é discutido à
luz de uma mudança de estratégia, de procura global e geral, para localizada.
Visual identification and detection processes show different sensitivity to the same context characteristics. In one case, context-congruent stimuli seem to be identified faster than incongruent stimuli; conversely, incongruency is detected faster than congruency. In both cases, the influence of expectations created from context is clear. Studies that use stereotypes as context have also shown that stereotype-congruent stimuli are identified faster. But little is known regarding the effect of these contexts in detection tasks. To test this effect, a flicker task was used in which 72 participants detected a changing object in natural scenes containing either a woman or a man. The object, the critical stimulus, was either stereotypically congruent (e.g., man-cigar) or incongruent (e.g., woman-tie). We expected differences between congruent and incongruent stimuli, such that changes in incongruent stimuli would be detected fast than changes in congruent stimuli. Results show that level of congruency did not impact detection speed. Performance was only affected by time between the start of a trial and occurrence of change (threshold): in trials when change happens at 1200ms, detection is significantly slower than in trials when it happens at 400ms or 800ms. We discuss these results within the body of knowledge concerning change detection and stereotypes, focusing both on methodologic problems with the task and the possibility that stereotypes might influence detection and identification tasks differently. The observed threshold effect is discussed within the framework of a change of strategy, between a gist-based global visual search, and local search.
Visual identification and detection processes show different sensitivity to the same context characteristics. In one case, context-congruent stimuli seem to be identified faster than incongruent stimuli; conversely, incongruency is detected faster than congruency. In both cases, the influence of expectations created from context is clear. Studies that use stereotypes as context have also shown that stereotype-congruent stimuli are identified faster. But little is known regarding the effect of these contexts in detection tasks. To test this effect, a flicker task was used in which 72 participants detected a changing object in natural scenes containing either a woman or a man. The object, the critical stimulus, was either stereotypically congruent (e.g., man-cigar) or incongruent (e.g., woman-tie). We expected differences between congruent and incongruent stimuli, such that changes in incongruent stimuli would be detected fast than changes in congruent stimuli. Results show that level of congruency did not impact detection speed. Performance was only affected by time between the start of a trial and occurrence of change (threshold): in trials when change happens at 1200ms, detection is significantly slower than in trials when it happens at 400ms or 800ms. We discuss these results within the body of knowledge concerning change detection and stereotypes, focusing both on methodologic problems with the task and the possibility that stereotypes might influence detection and identification tasks differently. The observed threshold effect is discussed within the framework of a change of strategy, between a gist-based global visual search, and local search.
Description
Dissertação de Mestrado apresentada no ISPA – Instituto Universitário para obtenção de grau de Mestre na especialidade em Psicologia Social e das Organizações
Keywords
Estereótipos Percepção Detecção Flicker task Stereotypes Perception Detection, Flicker task