Feature repetition effects on object familiarity: Evidence from an old/new recognition task

Feature repetition effects on object familiarity: Evidence from an old/new recognition task Selda Eren (e115275@metu.edu.tr) Department of Cognitive Science, Middle East Technical University Ankara, 06531 Turkey Annette Hohenberger (hohenberger@ii.metu.edu.tr) Department of Cognitive Science, Middle East Technical University Ankara, 06531 Turkey these feature domains (Hanna & Remington, 1996). Whenever a visual scene is encountered, activation is observed in these areas. Is this a mere bottom-up activation, or does the perceptual system attend to specific areas in the scene? We know that the visual system is not a passive receiver of visual data, but it actively obtains information from the visual flux (Jingling & Yeh, 2007). Attention makes a difference but we do not know whether the representation is stronger or the conscious access is easier in this case. Whether the features in the scene are selected or all stored, it is clear that a combination of these features constitutes visual representations (Slotnick, 2004). Also audial and tactile features can be integrated with visual features, in which case the resulting representation can be called an “event file” (Hommel, 1998). Hommel states that all the features perceived in the same temporal window are automatically stored in these event files. These files can include features of every type, blurring the distinction between different domains of features, including visual and spatial pathways, which are assumed to exist separately in the brain. He points to the importance of building arbitrary connections between the features from different domains for learning. In this study we investigated feature repetition effects on object familiarity. Hommel and Colzato (2009) report a decrease in performance in a stimulus-response task when one object feature is repeated while other features varied, as compared to complete repetitions and alterations. We predicted that repetition of particular features while other features vary would also affect familiarity of objects. We aimed to test this prediction with a continuous old/new study/test recognition design. In the study phase, participants saw a series of items one by one. In the test phase, they evaluated familiarity of the test items. To create the feature repetition effect, particular features were displayed more frequently than the other features in the study phase. We will call these features “frequently repeated features” (FRFs). In the test phase, items either had none, one, or two of the FRFs. We expected that the more FRFs the item had, the more participants would classify the item as familiar. We obtained scores for hits, misses, correct rejections and false alarms. False alarm scores are especially important Abstract We performed an old/new study/test recognition task to investigate feature repetition effects on object familiarity. The results showed that repeated features increased “old” responses during the test phase for new objects. This increase was linear with the number of repeated features on the object. Old objects, which had been among the study phase stimuli, were not affected by the number of repeated features on the object. We also analyzed the effect of feature type (colour, shape, border and pattern) on familiarity responses. We found an effect of feature type only for the old objects. Saliency of the features also affected familiarity: the more salient the repeated feature was, the more familiar the object was found. We propose that the feature repetition effect for the new objects might be due to (1) activation of more than one representation constructed during the study phase (2) a separate representation for the repeated features, which has the potential to interfere with several perceptual processes. Keywords: feature repetition effect; object recognition Introduction Formation and activation of perceptual representations has been the subject of various disciplines including, but not limited to, philosophy, psychology, psychophysics, neuroscience, and computer science. In philosophy, the existence of “mental representations” is a fundamental debate in the philosophy of mind. In psychology, the studies of categorization and memory directly relate to this problem. Artificial intelligence and robotics research concentrate on implementing visual systems that construct a representation of their virtual or real environments. With the emergence of cognitive science, the bodies of knowledge that developed in these separate fields are coming together, for a better understanding of how perceptual representations are constructed and accessed. This study aims to contribute to the research on the formation and activation of visual object representations by revealing some important factors involved in memory processes. Our approach takes its roots from findings in perception and memory literature and computational approaches in artificial intelligence. From a computational perspective, it is possible to represent everything in the environment as a combination of some features, like color, shape, pattern, etc. We know that the human brain has specialized areas for each of