The Primate Visual System

Jon H. Kaas, Christine E. Collins, Jon H. Kaas, Christine E. Collins

• Publisher: CRC Press
• Year: 2003
• eBook ISBN: 9781135513535
• Topic: Medicine
• Subtopic: Medical Theory, Practice & Reference

The last 20 years of research have been marked by exceptional progress in understanding the organization and functions of the primate visual system. This understanding has been based on the wide application of traditional and newly emerging methods for identifying the functionally significant subdivisions of the system, their interconnections, the

1 Parallel Visual Pathways in a Dynamic
SystemVivien A.Casagrande and David W.Royal

1.1 INTRODUCTION

The idea that information from different sensory modalities is processed in parallel can be traced to the 1800s when Johannes Müller put forth the “law of specific nerve energies.” ¹ The law in essence states that perceptions are determined by which nerve fibers are activated, not by how the nerve fibers are activated. For example, mechanical pressure to the eye produces a sensation of light, and activating axons in the auditory nerve by an electric shock gives rise to a sensation of sound. Today, we recognize that there are specific receptor cells, tuned to be sensitive to different forms of energy in the environment and that these receptor cells connect to specific nerves. Two other ideas about parallel processing of sensory information are well established. First, it is well accepted that sensory qualities within a modality, such as light touch vs. pain and temperature within the somatosensory system, are carried by separate, parallel pathways. This form of functional parallelism extends to other sensory systems including the visual system, the subject of this chapter. Second, within modalities, such as vision, audition, and somesthesis, information from different locations in the periphery is transmitted in parallel to the brain to maintain knowledge about spatial location. In other words, different locations on the skin, the cochlear membrane, and the retina send redundant signals about sensory qualities in parallel (topographic parallelism) to the brain to create maps of these sensory sheets. In the last case, the same sensory qualities are transmitted in parallel to allow for appreciation of these qualities at different spatial locations. Parallel processing of the type described implies that sensory experience is initially broken down into basic elements, which are transmitted in parallel, and that the reconstruction of the whole occurs at some central brain location. Historically, the idea was that there are sensory areas where separate senses are appreciated within the cortex and that these were later combined in an “association cortex.” The problem with this idea is that it now appears that most of the cortex is occupied by separate sensory areas or modules within areas (at least 32 visual areas have been recognized in macaque monkeys) leaving little room in cortex for association areas. ^{2 , 3} According to the current view, each higher order sensory area performs a separate specialized function or set of functions. It can be argued that parallel processing and modular specialization have the advantage that localized damage does not cause the entire system to malfunction. Also modular systems are easier to improve from an evolutionary standpoint since changes are not required within the entire network. Nevertheless, sensory modules need to receive input from somewhere and need to communicate their computational achievements to other parts of the system so the independence of these units can only be relative. Also, specialization is expensive because it requires dedicated units and there are not enough resources to have every sensation, thought, and action produced by separate cells, pathways, or modules. Additionally, effective behavior of the system as a whole requires smooth cooperation of components over very short time periods. Thus, the system must either be more integrated than it appears or have some means of tightly coordinating relevant tasks.

This chapter explores the question of parallel processing and the problem of integration in the primate visual system. Section 1.2 presents a brief history of parallel processing in the visual system showing how earlier views have channeled our thinking. Next, we consider how messages are defined within parallel channels and the degree to which these parallel channels beginning at the periphery are truly functionally specialized. We argue that the way messages are coded by channels is still a matter of debate and that separate channels likely evolve only under conditions where messages either are incompatible if carried by a single channel or result in loss of important information and that each channel carries more than one message. Section 1.4 considers the question of whether the signatures that define the parallel input channels from the lateral geniculate nucleus (LGN) can be traced to cells in primary visual cortex (i.e., V1 or striate cortex) or beyond this level to extrastriate visual areas. We argue that such LGN pathway signatures are difficult to recognize beyond the LGN and that V1 output pathways are not segregated according to the rules governing LGN parallel pathway inputs. Section 1.5 explores data that demonstrate that parallel visual channels carry information not just about vision but also about the other senses, as well as about eye movements and cognitive state. Section 1.6 discusses the dynamics and functional implications of parallel visual pathways. In particular, we explore the issue of differences in the timing of messages sent by different pathways and the impact of feedback on the messages that are sent. Timing clearly plays an important role and is critical for the subsequent integration of visual signals. Feedback also can alter feedforward messages. In these sections we also explore the differences and similarities between the organization of the visual system and other sensory systems as a way to uncover functional roles. The final section provides a summary and a list of unanswered questions about parallel visual system organization.

1.2
HOW ARE PATHWAYS DEFINED?

How we think about parallel processing in the visual system is a product of several distinct approaches to the problem and ways of conceptualizing brain function. At one end of the spectrum we can conceive of cells, pathways, modules, and areas in the brain as dedicated to one specific function. For example, the law of specific nerve energies implies that cells, pathways, and areas connected to the optic nerve will provide sensory qualities related to vision. At the other end of the spectrum are brain models that define functions through the activity of networks where cells contribute to a number of functions depending on which network is active (see References ⁴ and ⁵). The truth likely lies between the two extremes, as it is clear that in a basically segmented body plan like ours, neurons belonging to different segments specialize to perform different functions; however, these segments are not isolated but connected intimately to a larger network that coordinates purposeful behavior.

Discussed below are two key lines of investigation using different approaches that have strongly affected our views of parallel processing in the visual system. ⁶ In the first approach, parallel visual processing is treated as an engineering problem. In the mid-1960s, Enroth-Cugell and Robson ⁷ proposed that at its lowest level the visual system could work as a series of spatial filters, namely, as spatial frequency analyzers. In this model, cells tuned to different ranges of spatial frequencies respond to the appropriate frequency within the visual image and transmit this information centrally in parallel. Enroth-Cugell and Robson ⁷ used this linear systems approach to subdivide cat retinal ganglion cells into two types: those that summed luminance changes linearly across their receptive fields (referred to as X cells) and those that did not (referred to as Y cells). X cells were considered the “interesting” cells because they followed the logic of the model. This general approach led to numerous physiological and psychophysical studies based on the idea that the visual system’s response to any pattern could be predicted from its response to more basic temporal and spatial filtering components. In their original work, Enroth-Cugell and Robson, ⁷ however, did not argue that these cells limited their analysis to one spatial dimension or one attribute. They also described other properties that distinguished Y from X cells including the higher conduction velocities, sensitivity to higher speeds and lower contrasts, lower spatial frequency cutoffs, larger average receptive field center sizes, and more transient responses of Y vs. X cells. ⁷ These observations were important from the standpoint of parallel processing because they identified a collection of properties, not a single property (e.g., a single wavelength or single spatial frequency), that distinguished X from Y cells. Almost 20 years of studies on X and Y cells followed, showing that X and Y cells also could be distinguished based on morphology, retinal distribution, central targets, and receptive field properties (see Reference ⁸). From the constellation of traits defining each of these cell classes it was proposed that X cells were part of a channel to cortex subserving high-resolution pattern vision whereas Y cells were part of a channel subserving crude form and motion vision (see Reference 8). Also during this period, other cell types were discovered in the cat retina and LGN collectively referred to as W cells. The W cell category referred to those cells that investigators could not classify as either X or Y cells. Not surprisingly W cells were found to vary widely in properties, sharing in common only the attributes of low conduction velocity and relatively large receptive field sizes ⁸ (for review, see Reference ⁹). Because many W cells have heavy projections to the midbrain targets it was proposed that they subserve a more primitive (subcortical) type of vision referred to as “ambient vision.” X and Y cells, by contrast, provided “focal vision” or more highly evolved vision that required cortex. Ambient vision was seen as preconscious vision used by the earliest vertebrates to aid in spatial orientation and navigation relying on peripheral cues whereas focal vision was seen as the conscious, mostly foveal, vision used to identify and classify objects (the dominant form of vision in primates). The analysis of X, Y, and W cells in cats also led later to a similar set of investigations on LGN parvocellular (P), magnocellular (M), and koniocellular (K) cells in primates, where both similarities and differences between cats and primates were uncovered ⁶ (for review, see Reference ¹⁰).

The ambient/focal vision or the “two visual systems” hypothesis was actually linked to a second very influential set of studies begun in the 1960s by Gerald Schneider. Schneider published a key article ¹¹ in which he proposed that there was an anatomical separation between visual coding of the location (where) of a stimulus and its identification (what). Based on behavioral/lesion work in hamsters he argued that there were basically two visual systems supported by two separate pathways from the retina, the “where” pathway involving the superior colliculus and a “what” pathway involving the primary visual cortex (striate cortex or V1) ¹¹ (Figure 1.1A). The “where” vs. “what” or ambient vs. focal pathways were subsequently modified and described as independent pathways to separate cortical targets, one involving a pathway from colliculus to pulvinar to extrastriate cortex and the other from the LGN to V1 (see References ¹² and ¹³ and Figure 1.1B). The idea that these pathways were capable of independent parallel operation was demonstrated clearly in tree shrews where complete removal of V1 (and resulting complete degeneration of the LGN) or removal of the ...

Table of contents

• COVER PAGE
• TITLE PAGE
• COPYRIGHT PAGE
• SERIES PREFACE
• PREFACE
• EDITORS
• CONTRIBUTORS
• 1. PARALLEL VISUAL PATHWAYS IN A DYNAMIC SYSTEM
• 2. COMPARATIVE STUDY OF THE PRIMATE RETINA
• 3. THE PULVINAR COMPLEX
• 4. NORMAL AND ABNORMAL DEVELOPMENT OF THE NEURONAL RESPONSE PROPERTIES IN PRIMATE VISUAL CORTEX
• 5. MODULAR COMPLEXITY OF AREA V2 IN THE MACAQUE MONKEY
• 6. EARLY VISUAL AREAS: V1, V2, V3, DM, DL, AND MT
• 7. PLASTICITY OF VISUAL CORTEX IN ADULT PRIMATES
• 8. HIERARCHIES OF CORTICAL AREAS
• 9. VISUAL PROCESSING IN THE MACAQUE FRONTAL EYE FIELD
• 10. SPECIALIZATIONS OF THE HUMAN VISUAL SYSTEM THE MONKEY MODEL MEETS HUMAN REALITY
• 11. MAPS OF THE VISUAL FIELD IN THE CEREBRAL CORTEX OF PRIMATES: FUNCTIONAL ORGANIZATION AND SIGNIFICANCE
• 12. FACE EXPERTISE AND CATEGORY SPECIALIZATION IN THE HUMAN OCCIPITOTEMPORAL CORTEX
• 13. MOTION PROCESSING IN HUMAN VISUAL CORTEX
• 14. THE FUNCTIONAL ORGANIZATION OF MONKEY INFEROTEMPORAL CORTEX
• 15. COMPARATIVE STUDIES OF PYRAMIDAL NEURONS IN VISUAL CORTEX OF MONKEYS
• 16. FEEDBACK CONNECTIONS SPLITTING THE ARROW