人文科学研究所公開講演会開催のお知らせ

報告者1　　Simona Buetti 氏（イリノイ大学アーバナ・シャンペーン校心理学部特任准教授）

テーマ1　　 A new approach for estimating the contribution of different visual features to attentional guidance

要　旨1　　In this talk, I will present work that we have been doing in our laboratory to better measure and understand the contributions of different visual features to attentional guidance in visual search. We focus on visual searches that are visually-driven, that is, on visual searches where the visual distinctiveness of the target is sufficiently large that it can be processed in parallel, over the entire search display (i.e., efficient visual search). We begin by measuring the processing efficiency of a particular feature pair in a given feature dimension (i.e., how efficiently observers can find a target feature A among a field of distractors with feature B). And we do so for multiple feature pairs (e.g., find red among orange, red among green, etc…) and in different visual dimensions (e.g., color, shape, orientation, texture). We then create targets that differ from distractors along a combination of those features (color and shape, or shape and texture, or color, shape and texture). Our modeling demonstrates that the processing efficiency in these more complex (multidimensional) search conditions can be predicted to a great extent (Rsq>0.9) by the uni-dimensional estimates of processing efficiency. Further, one can also estimate the relative contribution of different feature dimensions to guidance. The methodological and theoretical implications of this approach will be discussed.

報告者2　　Alejandro Lleras 氏（イリノイ大学アーバナ・シャンペーン校心理学部教授）

テーマ2　　Towards a better understanding of the relationship between feature similarity and attentional guidance: the case of color
 

要　旨2　　Attention theories vary in how they implement the guidance of target information in visual search. Some theories propose that visual search is driven by similarity values (i.e., how perceptual similar the distractor feature is to the target feature), while others propose that visual search is guided by a tuned representation of the target color (i.e., the target feature is boosted or optimally tuned to favor the processing of the target feature in the display). Here, I present evidence that these two forms of guidance can be observed in color search, depending on the search task. First, we build models of what a similarity signal and tuned feature signal would predict. We then run a number of visual search experiments parametrically varying the target-distance perceived similarity in color space. We then use a model comparison approach to identify the conditions that lead to one versus the other form of guidance. Our results show that when participants are looking for a fixed target and the color of distractors vary from trial to trial, performance is best predicted (Rsq>0.96) by perceptual similarity. In contrast, when participants are looking for a fixed target and the color of distractors remains the same from trial to trial, performance is best predicted by a simple feature boosting model (Rsq>0.98), evidence that observers tuned their attention to the specific color discrimination they were judging on every trial. These results emphasize the importance of learning and trial history effects in determining how attention is guided by features in visual search.

使用言語：英語（通訳なし）

【注意事項】
・構内でのマスクの着用については個人の判断に委ねることが基本となりますが、　3密を回避できない場合はマスクの着用を推奨します。
・手洗い手指消毒などの基本的な感染防止対策にご協力ください。　　　　　　　　　　　　　　　

無料

※事前申し込みは不要です。

主　催　人文科学研究所

　　　　「認知心理学の応用」チーム（責任者：有賀　敦紀 研究員）