Cjep 61-3

We review the organization of the neural networks that underlie haptic object processing and compare that organization with the visual system. Haptic object processing is separated into at least two neural pathways, one for geometric properties or shape, and one for material properties, including texture. Like vision, haptic processing pathways are organized into a hierarchy of processing stages, with different stages represented by different brain areas. In addition, the haptic pathway for shape processing may be further subdivided into different streams for action and perception. These streams may be analogous to the action and perception streams of the visual system and represent two points of neural convergence for vision and haptics. Résumé We review the organization of the neural networks that underlie haptic object processing and compare that organization with the visual system. Haptic object processing is separated into at least two neural pathways, one for geometric properties or shape, and one for material properties, including texture. Like vision, haptic processing pathways are organized into a hierarchy of processing stages, with different stages represented by different brain areas. In addition, the haptic pathway for shape processing may be further subdivided into different streams for action and perception. These streams may be analogous to the action and perception streams of the visual system and represent two points of neural convergence for vision and haptics. Object recognition is a fundamental cognitive operation performed countless times each day. Yet despite decades of research into the mechanisms of human object recognition, we have only the barest idea of how this complex problem is solved so efficiently by the brain. Routine object recognition seems effortless and automatic to us, yet attempts to create artificial systems that recognize objects in the way that humans do have had little practical success. One suggestion for the slow progress of artificial recognition systems is the reliance of those systems on purely visual input, even though objects in our environment are a source of incredibly rich multisensory stimulation. For instance, a glass containing a soft drink can produce sensations of taste and smell, but you can also see the glass, watch the bubbles move, reach out and feel the bubbles burst against your skin and even hear them fizz. It is not a stretch to suggest that objects such as the soft drink are the rule, as opposed to the exception, in our world. It seems equally likely that when we are attempting to ascertain the identity of an object in our environment, we use all the information available, regardless of the sensory modality. However, despite the multisensory nature of real-world object recognition, until recently object recognition was almost exclusively studied using unisensory stimuli. Furthermore, the majority of those unisensory experiments used visual stimuli. Recently, though, there has been a surge of interest in multisensory phenomena, including multisensory object recognition (Calvert, Spence, & Stein, 2004). Because relatively less is known about how object recognition occurs using sensory inputs besides vision, the increased interest in multisensory recognition has led to increased interest in nonvisual unisensory object recognition. Of the various candidate sensory systems besides vision by which objects can be recognized, perhaps the most actively studied has been touch. Here, we will distinguish between passive touch and haptics, which we define as active use of the hands to retrieve the attributes of an object stimulus, using both cutaneous and kinesthetic inputs. Haptic object recognition has been studied behaviourally and in patients with brain damage for many decades. It has been studied using neurophysiologic and neuroimaging techniques since their inception. The intent of this chapter is to present an overview of the neural mechanisms of haptic object recognition. We will focus particularly on mechanisms involving the object attributes of shape and surface texture. Neural Mechanisms of Shape Recognition One self-imposed limitation on the breadth of object recognition research has been an emphasis on analyzing the geometric characteristics of objects (e.g., size or shape). Like the preferential study of vision over Thomas W. James, Sunah Kim, and Jerry S. Fisher, Indiana University Canadian Journal of Experimental Psychology, 2007, 61-3, 219-229 Canadian Journal of Experimental Psychology Copyright 2007 by the Canadian Psychological Association 2007, Vol. 61, No. 3, 219-229 DOI: 10.1037/cjep2007023 CJEP 61-3 8/27/07 9:29 PM Page 219