1ANALYSIS OF RETINOTOPIC MAPS IN EXTRASTRIATE CORTEXMartin I. Sereno, Colin T. McDonald, and John M. AllmanCognitive ScienceUniversity of California, San DiegoLa Jolla, CA 92093-0515andDivision of Biology 216-76California Institute of TechnologyPasadena, CA 92115Published as: Sereno, M.I., C.T. McDonald, and J.M. Allman (1994) Analysis of retinotopic maps in extrastriate cortex. Cerebral Cortex 4:601-620.AbstractTwo new techniques for analyzing retinotopic maps—arrow diagrams and visual field sign maps—are demonstrated with a large electrophysiological mapping data sets from owl monkey extrastriate visual cortex. An arrow diagram (vectors indicating receptive field centers placed at cortical coordinates) provides a more compact and understandable representation of retinotopy than does a standard receptive field chart (accompanied by a penetration map) or a double contour map (e.g., isoeccentricity and isopolar angle as a function of cortical x-y coordinates). None of these three representational techniques, however, make separate areas easily visible, especially in data sets containing numerous areas with partial, distorted representations of the visual hemifield. Therefore, we computed visual field sign maps (non-mirror-image versus mirror-image visual field representation) from the angle between the direction of the cortical gradient in receptive field eccentricity and the cortical gradient in receptive field angle for each small region of the cortex. Visual field sign is a local measure invariant to cortical map orientation and distortion but also to choice of receptive field coordinate system. To estimate the gradients, we first interpolated the eccentricity and polar angle data onto regular grids using a distance-weighted smoothing algorithm. The visual field sign technique provides a more objective method for using retinotopy to outline multiple visual areas. In order to accurately relate these arrow and visual field sign maps to architectonic features visualized in the stained, flattened cortex, we also developed a deformable template algorithm for warping the photograph-derived penetration map using the final observed location of a set of marking lesions.IntroductionOver half of the neocortex in primates consists of visual areas, many of which are retinotopically organized (reviews: Sereno and Allman, 1991; Kaas and Krubitzer, 1991; Felleman and Van Essen, 1991). The region of cortex involved, however, is large, and there is substantial variability among species and among individual animals of a species. Even well-defined areas like primary visual cortex (V1) and the2middle temporal areas (MT) show variability in their location, shape, size, appearance (see, e.g., Tootell et al., 1985 and Fiorani et al., 1989, on MT in New World monkeys). Since the majority of visual areas are not graced with the convenient suite of easily distinguishable features that characterize V1 and MT (e.g., prominent myeloarchitectonic borders, nearly complete, not-too-distorted, topological maps of the hemiretina), it has proved quite difficult to convincingly define their boundaries. Several different naming schemes have persisted for a number of the areas beyond V1, V2, and MT.Visual areas in the cortex are ideally defined on the basis of converging criteria (Allman and Kaas, 1971, 1974, 1975, 1976; Van Essen, 1985). These include visuotopic organization, architectonic features, connection patterns, and physiological properties. It is virtually impossible, however, to obtain detailed information about all of these criteria for multiple visual areas in a single animal. In the following set of papers, we have focussed on the first two criteria—visuotopy and architectonics. By concentrating on obtaining large numbers of recording sites in each animal, we have been able to much more clearly delineate the complex and variable mosaic of visual areas in primate dorsal, lateral, and ventral extrastriate cortex.In the present paper, we begin by describing our techniques for recording visual receptive fields from a large number of locations in each animal. In the process of analyzing these large data sets, it became clear that the standard technique for illustrating retinotopy—a numbered penetration site chart with correspondingly numbered receptive field charts—was much too unwieldy to handle multiple, partial visual field representations involving hundreds of sites. We needed to find a tractable way to view the entire data set for single animal, not the least, to avoid the temptation to extract the small portions of the data set that can often be found to support a simple story.We describe two new methods for representing large retinotopic mapping data sets in an interpretable way—arrow diagrams and visual field sign maps. We first illustrate conventional numbered receptive field/penetration plots and then show how arrow (vector field) diagrams make it possible to view the same data set much more compactly and understandably. Second, we describe a deterministic distance-weighted algorithm for interpolating receptive field eccentricity, angle, and diameter data onto regular grids so that we can generate isoeccentricity, isopolar angle, and isodiameter contour plots. Third, we describe how to make a visual field sign map (local mirror-image or non-mirror-image transformation of the visual field) from the interpolated isoeccentricity and isopolar angle grids. This technique brings out relations that are often subtle in an arrow diagram and completely opaque in a raw receptive field plot. Finally, we develop an iterative algorithm for warping an x-y penetration map derived from a recording photograph onto the stained, physically flattened cortex using marker lesions. This allows us to more accurately align our mapping data with anatomical landmarks.The overall picture of retinotopy in primate extrastriate visual cortex that we arrived at was substantially more complex than we had anticipated. The second companion paper (Sereno, McDonald, O’Dell, and Allman, forthcoming) uses the techniques developed here to analyze retinotopy and architectonic features of dorsolateral extrastriate cortical areas in the owl monkey. A detailed discussion of the 600+ point case illustrated here is contained in that paper. The third companion paper (Sereno, McDonald, and Allman, in preparation) examines ventrolateral extrastriate areas in the owl monkey. Portions of this work have