Sensorimotor neuroscience aims to understand planning and control of natural movements in the real world. However, experiments are performed in artificial environments specifically designed to answer a certain research question. To study the underlying neural mechanisms of a certain behavior, for instance straight reaches to different directions, researchers design experimental environments to isolate the behavior of interest. This bears the challenge to infer from experimental results to natural behavior for which no artificial behavioral constraints are applied. However, such isolation of a behavior is necessary to find the neural signals that correlate with the behavior of interest. For instance, if the gaze always follows an arm movement, do the recorded signal relate to the arm or eye movement? In a conventional experiment, researchers provide a clearly defined set of sensory inputs on which the subject is asked to perform a measurable action. Then they can analyze how the defined sensory input results in the measured behavior and draw conclusions of how the brain performs such tasks. To obtain further knowledge about computations performed by the brain, neuronal activity can be measured, for example by means of extracellular recordings in monkeys. In the end, researchers build models to interpret their observed correlation of sensory input, neural activity and behavior. However, for clear interpretations we need a sufficient knowledge about all three.
It is not the scope of this thesis to discuss the challenge to obtain appropriate neural signals, but it is equally important to understand the sensory input and the behavioral output generated by the brain (Krakauer et al., 2017). As a result, the sensory input, and alongside the experimental environment, is reduced to a necessary minimum without additional, potentially confounding, stimulations. The behavior is controlled by applying highly specific behavioral tasks, but also by means of physical restraints such as chin-rests, head fixation or arm fixation. For electrophysiological experiments with non-human primates, the monkeys are typically seated in a primate chair with only one hand having access to a manipulandum or touchscreen (Figure 1.4A). The chair imposes a fixed distance and orientation to a screen in an otherwise darkened room. Experiments in such environments lead to results clear enough to draw conclusions but bear the risk that we only witness a part of the picture too small for an appropriate interpretation.
The other extreme would be to let the monkey freely perform in an enriched environment (Figure 1.4B) without any instructions while monitoring behavior and environment with modern
1.3 From constraint towards freely moving non-human primates 13
A B
Figure 1.4:Chair seated vs freely moving monkey. A) Conventional monkey electrophysiological setup.
The animal is seated in a primate chair in a fixed distance to a screen and a manipulandum or touchpanel.
Often the head is fixed to the chair, partly as a requirement due to the tethered neural recording equipment.
B) With wireless technology, electrophysiological recordings are possible outside a conventional monkey chair. More complex behavior involving interaction with an enriched environment and whole-body movements can be studied. Modified from (D Foster et al., 2014) (CC-BY 3.0)
techniques such as motion capture (Ballesta et al., 2014; Nakamura et al., 2016). However, while a clear identification of complex behavior is anything but easy, and while neural recordings of freely moving primates is another challenge, we would not necessarily be able to interpret the data even if all technical issues were solved. First, it is possible to perform multiple movements at the same time. Without a clearly defined and known structure in the behavior it is difficult to identify which neuronal process relate to which behavior. For instance, the planning of a movement would likely occur during the execution of the preceding movement imposing a challenge to understand what neuronal processes relate to the executed or planned movement. Second, for a statistical analysis it is necessary to obtain repetitions of the investigated behavior, otherwise it is difficult to distinguish a meaningful neural signal from noise. For those two reasons, the research described later in this thesis expands the highly constraint experimental environments (to study reach movements), to a less constraining environment (to study walk-and-reach movements) while keeping the necessary control of behavioral and environmental parameters.
When working with monkeys or animals in general, a constraining environment raises animal welfare concerns. While monkeys are a seldom used model, they are a necessity for invasive studies in sensorimotor neuroscience due to their human-like ability to reach and grasp and their ability to solve complex cognitive tasks. Such tasks require intensive training by means of positive reinforcement training. During training or experiments, the monkeys are seated in a primate chair and divided from their social group. It is necessary to increase the incentive to engage with the task by applying a caloric or fluid control schedule, i.e. the monkeys obtain their daily food or fluid as a reward for successfully interacting with the behavioral task (Prescott et al., 2010). Several research groups interested in behavioral and cognitive research implemented
devices for training and cognitive testing of monkeys within their home environment for which monkeys perform hundreds of trials daily with ad lib access to fluid and food (Andrews &
Rosenblum, 1994; Bennett et al., 2016; Fagot & Bonté, 2010; Gazes et al., 2013; Kangas &
Bergman, 2012; Richardson et al., 1990; Washburn et al., 1989; Washburn & Rumbaugh, 1992).
The training approach presented by those studies differs in several respects to training chair-seated monkeys in sensorimotor neuroscience: 1) The monkeys can be exposed to the device for a longer period allowing them to choose their working regime in their own pace; 2) The monkeys can freely move in the cage sometimes even exposed to an enriched environment; 3) Often, the animals are in sight with their social group or even the whole group has access to such a device. Thus, a cage-based experimental setting has the potential to increase animal welfare relative to the conventional chair-based setting. Depending on the research question, it might not be beneficial to have a setup for which the animal is free to move or has access to its social group. However, it might still be possible, at least partly, to train the animals in a cage-based setting. Alternatively, such a setting could be used for preliminary tests to identify how individual animals cope with planned experiments. Such testing could be used to select animals for specific research projects.
In addition to the challenges of monitoring behavior and environment, neural recording techniques impose further constraints on movement. For instance, electroencephalography (EEG) easily picks up muscle activity which is stronger than brain activity, and even small movements in a scanner for magnetic resonance imaging (MRI) result in signal loss. In this thesis, I will focus on intracortical extracellular activity of individual neurons. This activity can be recorded from a microelectrode inside the brain. Conventionally, an electrode is inserted during the experiment through an opening in the skull by means of a micro-drive temporarily attached to the skull (Mountcastle et al., 1975). It allows searching for new neurons every session but requires head fixation, since the micro-drive would not withstand head movements. A more modern development are floating microelectrode arrays (Maynard et al., 1997; Musallam et al., 2007). Multiple electrodes are chronically implanted in the cortex only connected with a thin flexible cable to the electrical connector on the skull. While a readjustment of the electrode depth or position is not possible, it allows recording from many cells at the same time. And being fixed on the brain and not the skull, it is not susceptible to head movements.
However, monkeys are flexible animals that can easily reach the top of their head and climb on various structures. Even with floating microelectrode arrays, tethered neural recordings in a freely moving monkey are not possible or only under constraining circumstances (Ludvig et al., 2004; Sun et al., 2006). Recent technological advances led to wireless electrophysiological recordings in monkeys, and consequently, recording during unrestraint behavior (Agha et al., 2013; Fan et al., 2011; Fernandez-Leon et al., 2015; Grohrock et al., 1997; Jürgens & Hage, 2006; Miranda et al., 2010; Schwarz et al., 2014; Yin et al., 2014). A few studies already used such technology for studying freely moving monkeys in the context of locomotion (Capogrosso et al., 2016; D Foster et al., 2014), vocalization (Hage & Jurgens, 2006; Roy & Wang, 2012),