Fig. 9.1
Typical near-radial penetrations of cat SI. In this and the following figures illustrating histological sections, orientations of radial cords of cells are indicated by thin lines and recording sites are indicated by circles. Minimal RFs are drawn stacked in the order in which they were mapped in the penetration. COR coronal sulcus (From Favorov and Diamond 1990)
Minimal RFs mapped in arrays of closely spaced penetrations reveal the shape and size of macrocolumns delineated by such sharp minimal RF discontinuities (Favorov and Diamond 1990). An exemplary array of such penetrations is illustrated in Fig. 9.2. Twenty five penetrations were inserted in this experiment approximately perpendicular to the cortical surface and in each penetration two or three minimal RFs were mapped at or near layer 4 (Fig. 9.2a). Figure 9.2b shows the outlines of all 62 mapped minimal RFs on a single drawing of the skin of the forearm, revealing that they form 10 nonoverlapping clusters, labeled from a to j. Figure 9.2c shows a two-dimensional surface-view reconstruction of the cortical region sampled by the 25 penetrations. Positions of all the recording sites, projected radially onto layer 4, are indicated by filled circles. Recording sites mapped in the same penetration are connected by thin lines, and each recording site is assigned the letter of the RF cluster to which its minimal RF belongs. The reconstruction in Fig. 9.2c clearly shows that recording sites with the same minimal RF form nonoverlapping groups up to 350 μm in width, separated from other such groups by just a few tens of micrometers in the plane of cortical surface. The most probable arrangement of the columnar boundaries is indicated in Fig. 9.2c by the set of solid and dotted lines, revealing a mosaic of sharply delineated topographic units, within each of which the minimal RF does not change, but jumps to a new skin location when crossing into another such unit.
Fig. 9.2
Array of 25 closely spaced penetrations reveals discrete columns in cat SI (From Favorov and Diamond 1990)
To summarize, the minimal RF mapping method reveals sharp somatotopic discontinuities that subdivide SI cortex into a mosaic of 300–600 μm-diameter irregular hexagonally-shaped columns. The same minimal RF is mapped at any site within a column. In adjacent columns, minimal RFs occupy prominently displaced, nonoverlapping positions on the skin. The columns with more distally located minimal RFs tend to be larger in size. To emphasize their discrete nature, Favorov and colleagues called such place-defined macrocolumns “segregates.”
9.2.2 RF Diversity Within SI Macrocolumns
A minimal RF is determined jointly by multiple neurons in a vicinity of the electrode tip and is the skin site that provides the most effective afferent drive to those neurons as a group. Individual neurons in the recorded group obviously are driven from skin areas larger than their common minimal RF. For a single neuron, its maximal RF can be defined as the full extent of the skin area that provides suprathreshold input to that neuron (Favorov et al. 1987). Neurons located within a single macrocolumn/segregate all share in their maximal RFs that segregate’s minimal RF, but also each neuron receives afferent input from some additional – and frequently very extensive – surrounding territories, which are different for different neurons (Favorov et al. 1987; Favorov and Whitsel 1988ab).
Figure 9.3 offers an example of the diversity of maximal RFs of neurons found in a single segregate. Shown in Fig. 9.3a, 21 single neurons were isolated in a near-radial penetration of a single segregate in area 3b in cat and their maximal RFs were mapped. All 21 maximal RFs are drawn in Fig. 9.3b, revealing prominent diversity of their sizes, shapes, and skin positions across the length of the penetration and even among close neighbors. However, there is one very small skin locus that is common to all 21 RFs (Fig. 9.3c). This skin locus coincides with the minimal RF mapped in this penetration.
Fig. 9.3
Maximal RFs collected in a typical radial penetration of cat SI. Cortical locations of studied single units are indicated in panel A by tics (From Favorov and Diamond 1990)
To summarize, maximal RFs of all neurons in a segregate include a common skin locus (the minimal RF) and, in addition, extend for variable distances outward in all directions from that common locus. Neurons in a segregate differ from one another in how much their maximal RFs extend in each and every direction from the common skin locus (Favorov and Whitsel 1988b). Because of its central importance to segregate organization, the common skin locus of a segregate – its minimal RF – was given a special name, “segregate RF center” (Favorov and Whitsel 1988a).
9.2.3 Minicolumnar Organization of Maximal RFs
Mountcastle (1978) hypothesized that a radial cord of cells about 30–50 μm in diameter – a “minicolumn” – might be the smallest functional unit of neocortical organization. Structurally, minicolumns are attributable to the radially-oriented cords of neuronal cell bodies evident in Nissl-stained sections of cerebral cortex. Population analysis of maximal RFs of neurons isolated within the same SI segregate supports Mountcastle’s minicolumnar hypothesis (Favorov and Whitsel 1988a; Favorov and Diamond 1990). According to this analysis, the maximal RFs of neurons within minicolumns are most similar in size, shape, and position on the skin. In contrast, neurons located even in adjacent minicolumns typically have RFs that differ significantly in size and shape, and frequently overlap only minimally on the skin. In other words, local RF diversity within segregates is mostly attributable to diversity among minicolumns and much less to within minicolumns.
For example, Fig. 9.4 plots the average degree of overlap of maximal RFs as a function of the tangential distance separating neurons within a segregate. The plot shows that neurons that are the closest neighbors in the tangential plane of the cortex have the most similar RFs, and that similarity declines with increased distance. At separations larger than 50 μm, RF overlap within segregates is independent of the distance between neurons. Instead, moving from one minicolumn to the next within a segregate, maximal RFs shift back and forth on the skin without yielding a net RF shift across the entire segregate. Only at a border separating adjacent segregates do RFs shift en masse to a new skin territory (Favorov and Whitsel 1988a; Favorov and Diamond 1990).
Fig. 9.4
Similarity of maximal RFs as a function of distance separating two neurons in the plane of the cortical surface in monkey SI (From Favorov and Whitsel 1988a)
Because of the prominent differences in RF properties among neighboring minicolumns, even a point-like tactile stimulus ought to evoke a spatial pattern of activity in the responding SI region consisting of a mix of active and inactive minicolumns. High-resolution 2-deoxyglucose (2-DG) metabolic studies of monkey SI (Tommerdahl et al. 1993) indeed showed that column-shaped patches of 2-DG label evoked in SI cortex by natural skin stimuli comprise groupings of highly active minicolumns interdigitated with less active minicolumns.
9.3 Functional Significance of Discrete Place-Defined Macrocolumns
Mountcastle originally defined columns as functional entities comprising groups of minicolumns bound together by common input and short-range lateral connections. Since then, however, the term “column” has been frequently used more broadly to refer to any vertical cluster of cells that share the same tuning for any given RF attribute, not necessarily of any functional significance (Horton and Adams 2005). At the same time the functional significance of Mountcastle’s cortical columns has been questioned (Swindale 1990; Purves et al. 1992; Horton and Adams 2005; Da Costa and Martin 2010). For example, Horton and Adams (2005) conclude their comprehensive critique of Mountcastle’s columnar hypothesis by stating that “one must abandon the idea that columns are the basic functional entity of the cortex. It now seems doubtful that any single, transcendent principle endows the cerebral cortex with a modular structure. Each individual area is constructed differently…”
The minimal and maximal RF organization of SI segregates – Mountcastle’s original macrocolumns – however offers a number of insights that counteract the criticism and clarify the nature of macrocolumns as functional entities:
1.
The structural evidence of discreteness associated with macrocolumns should be sought at the level of cell bodies, rather than at the level of dendrites and axon terminals
The basal dendrites of pyramidal cells have a large horizontal spread – up to 400–500 μm in diameter (Feldman 1984). This means that a portion of the basal dendritic fields of the majority of pyramidal cells in an SI segregate invades neighboring segregates. If there were no restrictions on the inputs the dendrites of pyramidal and spiny stellate cells can receive from elements outside their own segregate, then the majority of the cells in a segregate would reflect the activity from the neighboring segregates invaded by their dendrites. Neurons in different parts of a given segregate would be influenced by activity of different surrounding segregates, and the RFs of a linear array of neurons across a segregate would show an orderly, gradual and continuous shift in skin position. In reality, segregates show no such somatotopic gradients within their confines (see above), thus indicating that although some dendritic branches of cells in one segregate invade the territories of adjacent segregates, they do not receive opportunistic synaptic contacts there from neurons residing there or from afferents innervating those segregates. On the other hand, the afferents innervating a given segregate and the local axon collaterals of neurons residing in that segregate will follow the dendritic branches originated in that segregate into the territories of the adjacent segregates. As a result, the systems of afferent and intrinsic connections wiring adjacent segregates can remain functionally separate while physically intermingled in each other’s neuropil.
2.
Definition of the macrocolumn by common input and short-range connections should not be taken to imply that neurons making up the macrocolumn all have uniform RFs
Instead of uniformity, neurons making up an SI segregate possess prominently diverse maximal RFs (see above), with each maximal RF differ in how much it extends in different directions away from the segregate RF center (i.e., the skin locus common to RFs of all the cells in the segregate). Thus, the segregate is recognized on statistical grounds by its possession of a certain assortment, or distribution, of maximal RFs centered on a particular skin point. Neurons in different sectors of the same segregate apparently have the same distribution of maximal RFs. The reason why the border between adjacent segregates can be detected is because RF distributions immediately across the border have clearly different central tendencies and occupy different, only partially overlapping skin territories (the RF center of one segregate is included in only 50 % of the maximal RFs found in an adjacent segregate). The reason why minimal RF mapping method is highly effective in detecting segregate borders is that it is designed to estimate the central tendency of the distribution of maximal RFs of a local group of neurons picked up by the tip of the recording electrode, and thus it directly reveals where in the course of an SI penetration the central tendency of local maximal RFs stays the same and where it jumps to a new skin location.
These statistical properties of segregate maximal RFs are depicted schematically in Fig. 9.5. It shows hypothetical distribution of maximal RFs across two adjacent segregates. For graphic clarity, the skin and RFs are treated as unidimensional. Hypothetical maximal RFs are plotted in Fig. 9.5b for 200 neurons sampled along a continuous path that spans the two segregates (Fig. 9.5a). The random variations in the size (length) of the RFs reflect the diversity of sizes of maximal RFs sampled in cat and monkey experiments (Favorov and Whitsel 1988a; Favorov and Diamond 1990). The first 100 RFs belong to neurons located in the first segregate. They all share in common skin point A, but otherwise they vary randomly how far they extend to the left and to the right from this central point of their distribution. The next 100 RFs belong to neurons located in the second segregate. Their distribution is similar to that of the first 100 RFs, except that it is centered on skin point B. The change in the central tendency occurs at the transition from the first segregate to the second, between the 100th and 101th RFs. The distance between the two central skin points, A and B, is such that the central point of one segregate is included in 50 % of RFs of the other segregate. This Fig. 9.5b plot captures the essential characteristics of the segregate organization of maximal RFs in cat and monkey SI cortex.
