Using VSDI in awake behaving monkeys, Slovin et al. (2002) imaged cortical dynamics with high spatial and temporal resolution from the same patch of cortex for up to 1 year (Fig. 11.4). In the original study, the visual cortices of trained macaques were stained 1–3 times a week. Immediately after each staining session, the monkey started to carry out a simple behavioural task, while the visual cortex was imaged using VSDI. The first important task of this study was to show that such repeated use of VSDI does not interfere with normal cortical function. The functional maps obtained by VSDI were confirmed by imaging based on intrinsic signals (Grinvald et al. 1986) over a period of up to a year (Fig. 11.4b). Orientation preference maps can be used as a sensitive assay to assess any potential cortical damage since orientation preference is formed in the cortex itself rather than lower brain areas. The imaging results, as well as the behavioural performance, indicated that pharmacological side-effects or photodynamic damage were negligible. Having established this point, Slovin et al. (2002) studied the spatiotemporal representation of a point stimulus in primary and secondary visual cortex (Fig. 11.4c), quantifying the latency and time course of the cortical response (Fig. 11.4d). The individual traces indicate that evoked activity can be detected without signal averaging in the awake monkey, offering VSDI as an important tool to explore real-time cortical dynamics. This is beneficial because averaging may mask important variability among the instantaneous responses. In an example behavioural experiment, the cortical dynamics as the monkey makes a saccade are depicted in Fig. 11.5. Evidently despite the high speed of the eye movements, an evoked response was detected also along the cortical representation of the fast eye movement.

With the improvements in the signal-to-noise ratio of in vivo VSDI offered by the blue dyes, evoked activity can be detected without signal averaging in the awake monkey (Fig. 11.4d; Slovin et al. 2002), as previously found in anaesthetized cats (Arieli et al. 1995, 1996; Tsodyks et al. 1999; Kenet et al. 2003). This finding indicates that cortical spatiotemporal dynamics per se can now be explored in awake behaving monkeys. Hebb (1949) suggested that neocortical neurons operate in assemblies, defined as networks of neurons, local or widespread, that communicate coherently to carry out the computations that are required for various behavioral tasks. Neurons located within the same pixel of a VSDI movie might belong to the same or different neuronal assemblies. It is therefore of great interest to monitor the spiking activity of a single neuron in the context of larger network dynamics, to reveal the co-active nodes in the assembly during a given computation. This can be achieved by combining single unit (or multiple-unit recording) with real-time optical VSD imaging of the dynamics of coherent neuronal assemblies. First, a single unit with the desired tuning properties is selected. Second, VSDI and unit recordings are conducted for a long time without any signal averaging. During off-line analysis, which involves performing spike-triggered averaging on the optical data, the firing of a single neuron serves as a time reference to selectively visualize only the population activity that was synchronized with it. With a sufficient number of spikes, any neuronal activity not time-locked to these spikes will be averaged out and the net, clean spatiotemporal pattern of coherent activity will be thus obtained.

Such spike-triggered averaging of VSD signals can be performed on both evoked and spontaneous activity. Even in primary sensory areas, there is a large amount of spontaneous on-going electrical activity, primarily subthreshold activity, in the absence of any sensory input. Some of this activity is coherent over large cortical areas. Is this spontaneous activity just network noise? How large is it relative to evoked activity? Does it affect evoked activity? And does it have an important functional role? VSDI is ideally suited to exploring such questions at the level of neuronal assemblies. To compare the amplitude of coherent, ongoing activity with that of evoked activity (both representing large subthreshold activity, to which VSDI is sensitive), spike-triggered averaging was used during both spontaneous- and evoked-activity imaging sessions with anaesthetized cats (Grinvald et al. 1991; Arieli et al. 1995, 1996; Tsodyks et al. 1999; Kenet et al. 2003). Surprisingly, the amplitude of coherent, spontaneous, ongoing activity in neuronal assemblies was nearly as large as that of evoked activity. Large, coherent, ongoing activity was also found in the somatosensory cortex (Petersen et al. 2003b; Ferezou et al. 2006, 2007). The large fluctuations of ongoing activity affect how far cortical neurons are from their firing threshold. Therefore, the amplitude of ongoing activity indicates that it might have an important role in shaping spatiotemporal patterns evoked by sensory input. The idea that network activity can converge to cortical states is a central concept in theoretical brain research. What are the dynamics of intrinsic cortical states? It is now possible to visualize certain aspects of cortical networks and their states in action, at a high spatiotemporal resolution. Such studies can provide insights into the dynamic interplay between activated internal cortical representations and incoming sensory input. The previous studies on anesthetized cats are reviewed in the previous chapter. In those VSDI studies we reported that spontaneous ongoing cortical activity in areas 17 and 18 represented dynamic spatial patterns, and about 20 % of these patterns resembled the functional architecture of orientation domains. These patterns covered large cortical areas—up to the entire imaged areas of up to 6 × 6 mm (Grinvald et al. 1989; Arieli et al. 1995, 1996; Tsodyks et al. 1999; Kenet et al. 2003).

To find whether these results are relevant to the awake behaving primate, Omer and Grinvald (2008) performed VSDI of ongoing cortical activity in the visual cortices of awake monkeys simultaneously with measurements of single unit activity and the local-field potential. They found coherent activity also in the awake monkey (Fig. 11.6). This figure shows the tuning properties of the reference cell (Fig. 11.6a), the border between the two imaged areas V1 and V2 (Fig. 11.6b), time course of the coherent activity in area V1 (red) and V2 (blue; Fig. 11.6c) and the instantaneous spatial pattern of the assembly activity just before the action potential of the reference neuron (Fig. 11.6d). Overall, the dynamics were very different from that found in anesthetized cats and in a single epoch it was not possible to recognize any pattern that looked like the known functional architecture. Nevertheless, it was found that such patterns do appear: the pairwise correlation among pixels of known functional architecture as a function of the orientation difference between these pixels or their ocularity index had the same stereotypic shape for evoked activity and for spontaneous epochs. Similar results were obtained in anesthetized monkeys. However, in the anesthetized monkeys, Omer and Grinvald found that spontaneous cortical activity resembled the functional architecture just like in the anesthetized cat (unpublished). These results underscore the importance of carrying out experiments on awake behaving animals since the spatiotemporal patterns of activity are so drastically different.

Microstimulation in various brain regions has been used extensively to affect neuronal and behavioural responses. However, many questions remain about the spatiotemporal patterns of activity that are evoked by microstimulation and their characteristics as a function of various parameters such as the current strength and the electrode shape (Tehovnik et al. 2006).

Intracortical stimulation has been used as a tool to discover the pattern of muscle and/or movement representation in motor areas of the cortex (Slovin et al. 2003). The direct and synaptic spread of activation induced by stimulation has been the subject of considerable uncertainty and controversy. To help resolve some of these issues we used voltage sensitive-dye (VSD) imaging to examine the patterns of activation evoked by different stimulation parameters in the primary motor cortex (M1) of an awake monkey. Single cathodal pulses in M1 (15–30 μA, 0.2 ms duration) evoked short latency VSD responses that continued for 40–50 ms after the onset of stimulation. These responses spread in an asymmetric manner from the stimulation site in M1; the width of the activated area was about 1.5–3 mm. Single pulses larger than 15 μA also evoked activity at spatially separate sites within M1. Short trains of pulses in M1 (2–20 pulses, 330–500 Hz, 30–100 μA) produced activation that spread from the electrode site to adjacent premotor areas and to the contralateral hemisphere. A distinct hyperpolarization followed the activation evoked by a high current (>60 μA), short stimulus train (2–20 pulses). Finally, long stimulus trains (70 μA, 200 pulses, 330 Hz) evoked activation that spread over large portions of M1 and the adjacent premotor areas. These responses outlasted the stimulus train by several hundred milliseconds. Our results emphasize the importance of considering the spread of activation when evaluating the results of M1 stimulation.

VSDI studies on the effect of microstimulation in the frontal and motor cortices of awake monkeys have also recently shed light on the evoked spatiotemporal dynamics. The frontal eye field and neighbouring area ‘8Ar’ of the primate cortex are involved in the programming and execution of fast eye movements. Electrical microstimulation in these regions elicits short-latency contralateral saccades. A combination of VSDI and microstimulation has been used to determine how the spatiotemporal dynamics of microstimulation-evoked activity are converted into saccade plans. Microstimulation was shown to elicit neural activity with complex spatiotemporal dynamics, both inhibitory and excitatory; these dynamics depend on the stimulated area, and have important behavioural correlates. The observed spread was large and depended on the amplitude of the microstimulation (Seidemann et al. 2002). These results emphasize the importance of further characterization of microstimulation-evoked activity for the interpretation of its behavioural effects and what it has revealed about the neural basis of cognition (Newsome et al. 1989; Cohen and Newsome 2004). VSDI seems to be one of the most appropriate tools to nail down these issues, which also have important implications for neural prosthesis and brain machine interfaces.