Pyramidal neurons are distinguished by their prominent apical dendrite and basal dendritic tree (Fig. 1.3). They comprise some 70–90 % of all neurons in cerebral neocortex (DeFelipe and Fariñas 1992). Pyramidal cells form rich plexuses of connections, often forming intrinsic lattices or patches, within cortical areas. They form nearly all cortico-cortical connections, both ipsi- and contralateral, as well as most subcortical connections. Pyramidal cells contain the excitatory neurotransmitter glutamate: their discharge directly facilitates cortical activity, rather than inhibiting it. Arguably, pyramidal cells are the principal neurons of the cerebral cortex, generating nearly all cortically initiated excitation. Different subtypes of pyramidal cells have been reported based on aspects of their morphology, for example, inverted pyramidal cells, those whose axons project into the white matter, and those whose axons are restricted to the grey matter (see Feldman 1984; Nieuwenhuys 1994; Elston and DeFelipe 2002; Valverde 2002 for reviews). In addition, pyramidal cells have been subdivided according to their neurochemical content and receptor subunit profiles (Gabernet et al. 1999; González-Albo et al. 2001; Hof et al. 2001). It has also been proposed that pyramidal cells are not genetically fated to have their characteristic morphology. According to this theory, all variants of spiny neurons, from the typical pyramidal cell to the typical spiny stellate cell, are derived from a common precursor (Valverde 1988). Here we focus on the “typical” pyramidal cell of unknown neurochemical content observed in mature primate cortex. As a prelude to studying specialisations in pyramidal cell structure that have occurred during the evolution of different primate species we first set out how pyramidal cell structure varies among Brodmann’s areas in a single species of macaque monkey (Macaca fasicularis), and outline how specialisation in cell structure may influence functional capabilities and, in turn, behavioural complexity.

The areas of the cerebral cortex that contain neurons involved in some form of visual processing are perhaps the most thoroughly explored in the macaque brain. Prior to the 1970s, most studies were restricted to Brodmann’s area 17 (the striate, or primary (V1), visual cortex). Since then there has been an explosion in the number of studies in, and our understanding of, extrastriate visual cortex (Zeki 1969, 1978a) see (Kaas 1995; Rosa 1997; Kaas and Lyon 2001; Kaas and Preuss 2003; Zeki 2003; Rosa and Manger 2005) for reviews. Application of techniques such as electrophysiological mapping and imaging, and the development of specialised tracers, have revealed that visual processing is much more complex than previously thought, involving up to half the cortex and as many as 30 different areas (Fig. 1.3). Various theories have been presented for the existence of so many visual cortical areas, and how visual stimuli are processed by neurons in these areas (see Kaas 1987; Weller 1988; Felleman and Van Essen 1991; Rosa 1997; Kaas 2000 for reviews). In addition, many theories have been proposed regarding the recruitment and interaction of neurons in these different cortical areas during particular visual tasks. Two of the most popular models include the quasi-hierarchical model and the distributed processing model see (Mountcastle 1995) for a review. In the quasi-hierarchical model, visual inputs to cortex are processed through a series of cortical areas. These areas are not necessarily organised into a strict hierarchy, but there is some form of serial processing through select visual areas. In the distributed processing model, visual processing occurs in multiple cortical areas, but not necessarily in any form of hierarchy. That is not to say however, that the two theories are mutually exclusive. Mountcastle (1995), for example, highlighted how hierarchies may exist within a distributed system.

New methods of quantification (Elston and Rosa 1997; Elston 2001) have revealed marked, systematic differences in pyramidal cell structure (and cortical circuitry) in these different visual areas. Briefly, there is a trend for increasingly more complex pyramidal cells with progression from V1 to the second visual area (V2) and parietal visual areas (the lateral intraparietal area, LIP, and cytoarchitectonic area 7a), and temporal areas (V4, the middle temporal area, MT, cytoarchitectonic areas TEO and TE, and the superior temporal polysensory area, STP) (Elston and Rosa 1997, 2000; Elston et al. 1999a). The increase in the size of the dendritic tree, coupled with a concomitant increase in the number of dendritic branches and spine density, results in a progressive doubling in our estimates of the total number of spines in the basal dendritic trees of pyramidal cells through V1, V2, V4, TEO and TE. The functional implications of these specialisations in pyramidal cell structure in functionally related cortical areas are discussed in detail in the works of Elston (2002, 2007), Jacobs and Scheibel (2002), Spruston (2008) and DeFelipe (2011).

Based on patterns of connectivity, neuronal response properties and, more recently, imaging studies, several theories have been put forward regarding normal function across, and cooperation between, Brodmann’s sensorimotor areas (Mishkin 1979; Pons et al. 1987, 1992; Passingham 1997; Geyer et al. 2000). By injecting large numbers of pyramidal cells in some of these different cortical areas it has been possible to demonstrate marked and systematic differences in their size, branching complexity and spine density. More specifically, two trends of increasing morphological complexity have been revealed with progression from the central sulcus to adjacent cortical areas. There is a systematic increase in the size of pyramidal dendritic trees, their branching complexity, and spine density within their basal dendritic trees, with caudal progression from the primary somatosensory area on the posterior wall of the central sulcus (Brodmann’s area 3; 3b) to the rostral bank of the intraparietal sulcus (Brodmann’s area 5) and the exposed rostral portion of the inferior parietal lobule (Brodmann’s area 7; 7b). There is also an increase in the size of the dendritic trees of pyramidal cells, their branching complexity, and the total number of spines within their basal dendritic trees, with rostral progression from the primary motor area on the anterior wall of the central sulcus (Brodmann’s area 4) to the exposed lateral portion of the precentral gyrus (Brodmann’s area 6 or premotor cortex) (Elston and Rockland 2002). These differences in size, branching complexity and spine density result in appreciable differences in our estimates of the total number of spines in the basal dendritic tree of the average cell in each cortical area (Fig. 1.3).

A study of the literature reveals little agreement regarding the functions performed in cingulate cortex. Some have attributed higher cognitive and emotional functions to the anterior cingulate cortex and vegetative functions to posterior cingulate cortex (Allman et al. 2001) whereas others have claimed the reverse (eg Baleydier and Mauguiere 1980). In a series of studies in which cortical activity was recorded in awake behaving monkeys by functional magnetic resonance imaging (fMRI), Dreher and colleagues revealed that neurons in anterior cingulate cortex, unlike those in posterior cingulate cortex, are often co-activated with granular prefrontal cortex (gPFC) during cognitive tasks (Dreher and Berman 2002; Dreher and Grafman 2003). Various other studies have also demonstrated “executive” or cognitive function in the anterior cingulate, including error detection and reward-based decision-making (Gemba et al. 1986; Carter et al. 1998; Bush et al. 2002; Shidara and Richmond 2002; Hadland et al. 2003).

While it is well known that these two regions of cingulate cortex can be distinguished by their cytoarchitecture, particularly the granular layer, relatively little is known of possible differences in their microcircuitry. In order to investigate this, layer III pyramidal cells were injected in the posterior cingulate gyrus (Brodmann’s area 23) and their structure compared with that of cells injected in the anterior cingulate gyrus (Brodmann’s area 24) (Elston et al. 2005a). These investigations revealed that pyramidal cells in area 24 were considerably larger, more branched and more spiny than those in area 23. Estimates of the total number of spines in the basal dendritic tree of the average cell reveal a two-fold difference between cells in areas 23 and 24 (Fig. 1.3). Moreover, pyramidal cell structure in area 24 more closely approximates that of cells in the gPFC, than does that of cells in area 23 (see below).

Exactly what constitutes prefrontal cortex (PFC) has been interpreted in many ways. Some classify it as cortex that receives projections from the medial dorsal nucleus of the thalamus whereas others distinguish the PFC by cytoarchitecture (see Fuster 1997 for a review). Here, as intended by Brodmann (1913), the term is used to only include granular cortex anterior to the central sulcus (see Elston and Garey 2004). To avoid confusion here this region is referred to as the gPFC. Brodmann’s original maps of the gPFC have been further refined by others by cytoarchitecture and patterns of corticocortical connectivity (Vogt and Vogt 1919; Walker 1940; Barbas and Pandya 1989; Petrides 1991; Preuss and Goldman-Rakic 1991a, b, 1991c; Petrides 1998; Pandya and Yeterian 2000; Petrides and Pandya 2001). Broadly speaking, these different areas have been grouped into the lateral, medial and orbital regions.

Prefrontal cortex has been the focus of intensive investigation in recent times because of its involvement in executive functions such as conceptual thinking, prioritising and planning (see Goldman-Rakic 1996; Fuster 1997; Barbas 2000; Petrides 2000; Miller and Cohen 2001 for reviews). Our understanding of the functions performed by neurons in the different regions within gPFC, and how other cortical regions participate in specific tasks, is growing rapidly with the advent of new methodologies (see Quintana and Fuster 1999; Passingham et al. 2000; Rolls 2000; Fuster 2001; Funahashi and Takeda 2002 for reviews). Orbital and medial gPFC are now believed to be involved in emotional behaviour and the processing of taste, reward, memory, affect and motivation. The lateral gPFC provides cognitive support to the temporal organisation of behaviour, speech and reasoning and is involved in executive control of voluntary motor movements. Caudo-rostral gradients characterised by different patterns of connectivity and functions have also been reported within each of these gross subdivisions (Goldman-Rakic 1987; Petrides 1987, 1991; Barbas 1992; Wilson et al. 1993; Barbas et al. 1999; Pandya and Yeterian 2000). What is clear from these studies is that because of the complexity of functions performed by neurons in gPFC, and the difficulty in quantifying neuronal responses to specific executive tasks, our understanding of prefrontal function is likely to lag behind that of sensory cortex for some time to come.