Purpose of the Study

The changes in activity of single neurons and the local field potential (LFP) in area 18 under similar ex- perimental conditions as this dissertation work have been investigated previously (Galuske et al., 2002; Ebisch, 2007), partly based on the same dataset as for the present dissertation (Geider, 2008; Barnes, 2014). However, neurons are neither isolated entities, nor does the compound activity of many neurons (as observed with methods with lower spatial resolution, such as EEG and LFP) occur in isolation of the activity in distant cortical locations. Rather, local neural activity is integrated in a complex global net- work, and only through concerted activity of this global network can complex tasks be solved. Therefore, the analysis of local functional networks performed in this work can provide important contributions to the resolution of research questions in neuronal signal processing and "close the gap" between single-cell and LFP analysis.

This thesis therefore starts off with the challenge of assessing the properties of local neuronal net- works in a dataset recorded in cat area 18, while the posterior middle suprasylvian (pMS) cortex is thermally deactivated. The cells in deactivated pMS cortex, and hence the feedback signals to area 18, are thus silenced. This is assumed to alter the activity patterns in the recorded region, area 18.

In order to assess changes in network structure during the deactivation phases, methods that help define the links in the area 18 network are required, as well as methods that are able to disclose the network properties. Aspects that should be taken into account during such analyses are the suitability of the applied connectivity measure, the time windows used for correlation measures, the definition of the nodes of the networks, and the relevant spatial scales.

The goal of this study is the characterization of the functional properties of the recorded neurons under different experimental conditions. Particularly, the changes in interactions between (small groups of) cells are described. Spike rates, orientation and direction selectivity, and correlated activity among neurons are assessed, which are eventually used to create networks based on significantly correlated

activity. To this end, methods and tests were developed, and used to ascertain and interrogate local functional networks.

The focus of this study will, however, not only be on the methods used and (further) developed through application to the datasets described below, but also on the interpretation of the results of the network analysis, as they pertain to the neurophysiology. With this, there are three levels of interest: 1) the changes in spiking activity and local network properties in area 18 during deactivation of pMS, 2) a more general characterization of the firing properties and network connectivity patterns of cells in V1/area 18, and 3) the even more general question of the principles of neuronal coding in the visual cortex, and the neocortex in general.

2 Material and Methods

In this chapter, the experimental and analytical methods that were used in the study are elucidated.

2.1 Experimental Procedures

The analysis methods described below were applied to a dataset obtained in a series of experiments con- cerning visual neglect. The analyses are based on electrophysiological action potential and local field potential recordings. The experiments were conducted at the Max Planck Institute for Brain Research, Frankfurt, Germany. All procedures were performed following the guidelines of the local authorities and overseen by a veterinarian.

Please note that the author did not perform the deactivation experiments, but was provided with the raw datasets after the recordings. Experiments were conducted by Boris Ebisch, Kirsten Geider, William Barnes, and Ralf Galuske. The same dataset has been analyzed with a different focus by Kirsten Gei- der (2008) and William Barnes (2014). These publications, as well as the doctoral thesis by Boris Ebisch (2007), also provide further details on the experimental procedures. Here, only a broad overview shall be given.

Data were collected in three anaesthetized male cats aged one and a half years, using an Eckhorn multi-electrode system (Thomas Recording, Germany) (Fig. 2.1). Sixteen custom-made tungsten elec- trodes were placed in area 18 of the cat’s primary visual cortex, with a distance of 500µm in the x-y- plane, but individually adjustable in cortical depth. Cooling probes, also known as cryoloops (Lomber et al., 1999), were implanted in the pMS cortex to induce a visual hemineglect (Payne et al., 1996).

Figure 2.1.: Electrode matrix. The electrodes were placed in a 4x4 grid in a so-called "Eckhorn matrix", with an equal spacing of 500 µm in the 2D-plane, but each electrode was individually adjustable in depth with the aid of micro motors (photo by Matthias Munk).

Figure 2.2.: Deactivation conditions. There were three possible deactivation conditions, as depicted in this gure. The deactivated area is shown in black, i.e. ipsilateral and bilateral deactivation is shown. Recordings were performed in the ipsilateral and contralateral area 18 on both hemispheres. Bilateral recordings were not performed. Modied from Lomber et al. (1999).

Electrophysiological signals were recorded in area 18, part of the primary visual cortex of the cat, while the ipsilateral, contralateral, or both pMS cortices were thermally deactivated (Fig. 2.2). For the deactivation, chilled methanol was pumped through the cryoloops. Figure 2.3 shows a schematic of the experimental setup.

Stimulation and recording protocol

The animal was visually stimulated with a black and white contrast square-wave grating, which was viewed by both eyes simultaneously. The grating was presented in four different orientations (0◦, 45◦, 90◦, 135◦), which moved in either of the two directions perpendicular to the bars of the gra- ting. Hence, eight different stimulus conditions were possible (Fig. 2.4).

The order of occurence of the stimulus conditions was pseudo-randomized. Visual stimulation started with showing a grey screen for two seconds followed by one of the four differently oriented gratings remaining stationary for two seconds before the grating started moving for four seconds. Thus, one trial lasted for eight seconds.

Anaesthesia

Anaesthesia was induced by intramuscular administration of ketaminehydrochloride (ketamine 10%, 10 mg/kg body weight), xylacinehydrochloride (1 mg/kg body weight), and atropinsulfate (0,1 mg/kg body weight). For maintenance of anaesthesia, animals were artificially ventilated using a mixture of N2O (ca.

70%), O2(ca. 30%), and halothane (1-1.2%). The concentration of O2, CO2, and halothane in inspiration

and expiration was monitored closely throughout the experiment.

Multi-unit spike signal

In order to obtain spiking data, the signal was recorded with a sampling rate of 20-22 kHz and band- pass filtered (800-5000 Hz). Details on the analysis of the spike data are given below.

Local Field Potential

Parallel to the recording of the spike signal, a low-pass filter (120 Hz) was applied to the signal to obtain local field potential (LFP). This signal was digitized using a sampling rate of 1 kHZ. A Fast Fourier Transform (FFT) was applied to obtain the spectral composition of the signal. The power for different frequency bands between 4-90 Hz (see table 1.1, p. 16) was computed separately for further

Electrode Matrix

Pump

Thermo-

meter

Cooling

loop

Figure 2.3.: Cooling circuit. The gure shows the position of electrodes and cryoloops in the cat's brain. Tungsten electrodes, operated using micro motors, were unilaterally placed inside area 18, while cooling probes were bilaterally placed into the pMS sulci. Methanol was cooled with dry ice and pumped through the circuit. The temperature at the cooling site was monitored and controlled by changing the velocity of the methanol ow. (Brain modied from Tusa et al. (1981); Galuske et al. (2002)1_{, cooling circuit} modied from Lomber et al. (1999), photo of electrode matrix by Matthias Munk.)

Figure 2.4.: Stimuli. A black and white square-wave grating was shown in four dierent orientations, and eight dierent moving directions.

analyis. With this, oscillations with a frequency of 50 Hz, i.e. of the European line noise, were omitted. Evoked and induced activity were calculated by computing the spectral composition of each trial and then averaging (induced activity), or averaging over all trials first and then applying the FFT (evoked activity), respectively.