David Crow
david.crow@ a c m . o r g

Section A
Perception - Behavioral with neuroscience techniques

The understanding of perception has been advanced by combining purely behavioral with neuroscience techniques. Give 3 specific examples of research areas where this joint approach has been beneficially applied, and identify the way that behavioral and neuroscience findings have dovetailed.

1. Visual motor system - Milner and Goodale (in Kolb & Whishaw, 1996)

The visual system has been studied in great depth by both behaviorists and by neuroscientists. The neuroscience approach has defined the visual pathways from the retina to the cortex. Ungerleider-Mishkin have proposed a model of visual processing beyond the occipital lobe. First, they suggest that there is a dorsal stream that goes from V1 to the parietal areas of the cortex that is used in the perception of the size, shape and location of objects. Secondly, they suggest a ventral stream that goes from V1 to the temporal visual areas of the cortex that is associated with object recognition. There is a great deal of behavioral and anatomical evidence that supports this distinction. Monkeys with parietal lobe lesions have difficulty in directing their eye movements and their limbs in space. Similarly, there is evidence that damage to the temporal lobe results in failures of visual recognition and visual learning.

Milner & Goodale suggest that the behavioral evidence supports the dorsal stream to be used for location, size, shape and visually guided movements. This proposal stems from a patient who was blind, but her dorsal stream was intact as evidenced by her being able to see "unconsciously" location, shape and size. This patient could also shape her hand appropriately when reaching for an object. Further evidence of the dorsal stream being a system for visual control of action is from monkeys. Certain cells may be active when a monkey reaches out to a particular object, just looking at an object does not activate these neurons. This suggest that this system is used also when the brain is acting on the visual information. This has lead to the development of an understanding of the pathways of the visual cortex. It is thought that there is 3 distinct pathways, from V1 to V5 to the parietal areas (motion), from V1 to VP and V3A to the parietal regions (form), and from V1 to V2 to the parietal cortex. This model has provided a different way of describing the functional division between the dorsal and ventral streams of visual processing.

2. Connectionist models of word perception

McClelland and Rumelhart (1981, in Anderson, 1995) proposed a model for the combination of stimulus and context information in pattern recognition. Specifically the model uses word structure to facilitate the recognition of individual letters. This model is a connectionist representation of a pattern-recognition network. Individual features, like a vertical bar or a semicircle, are combined to form letters and individual letters are combined to form words. The activation of the individual letter features, which are very similar to the patterns detected by the retina, spreads from the letters and from the letters to the words. Alternative letters and words inhibit each other. The activation from the words can spread down to the letters composing the word and hence promote the recognition of the word.

The interactive activation model described above is a model that simulates simple neural processing. Behavioral data has been used to increase the validity of this model. Specifically, Massaro (1989) argued that the interactive activation model "was to insensitive to effects of stimulus information that contradicted the information from the context" (Anderson, 1995, pp. 70). When the PDP (TRACE model proposed by McClelland & Elman, 1986) model’s performance was compared to subject data on the performance in a phoneme recognition task it showed much stronger effects of context than shown in the data. This model was adjusted by McClelland (1991, in Anderson, 1995) by using activations that were somewhat variable. This model does a better job a predicting how subjects combine stimulus and context information. The use of behavioral data to improve PDP models is one example of how behavioral and neuroscience approaches can be used to enhance the understanding of each theory and model.

3. Imagery and neuro-imaging

Visual imagery has been thought to be very similar to visual perception but until recently there was only behavioral evidence to support this claim. Such evidence includes the work of Sheppard and his colleagues on mental rotation, the work of Kosslyn et al. on image scanning. This work suggests that the amount of time required to determine if two objects are identical is a linear function of the number of degrees the image must be rotated. And Kosslyn et al. found the amount of time required to scan between two objects in a mental image is a function of the distance between the objects. This provides support that the manipulations of mental images is a linear function of the manipulations that is performed on the actual objects. There remains a question of if the brain is using the same system for the processing of visual images as for visual perception. In 1993, Kosslyn, Alpert, Thompson, Maljkovic, Weise, Chabris, Hamilton, Rauch, and Buonanno took a different approach to the comparison of imagery and perception (in Anderson, 1995). They has subjects view blocks of letters or image them. They measured activity of the visual cortex using PET scanning technique. They found that there was greater activity of the visual cortex during imaging than during visual perception. They speculate that the activity was greater because imagery requires more effort than the visual representation than does perception. This approach has shown that visual images share many properties with the products of visual perception including the areas of the cortex that are used when processing either type of information.

Anderson, J. R. (1995). Cognitive Psychology and its Implications (4th ed.). W.H. Freeman and Company, New York.
Kolb, B &Whishaw, I. (1996) Fundamentals of Human Neuropsychology. Chapter 11, Occipital lobes. p243-264.