A New Experimental and Conceptual Approach to Understanding the Ventral Tegmental Area and Its Regulation of Motivated Behaviors
Motivated behaviors are essential for the survival and maintenance of life. The ventral tegmental area (VTA) is a midbrain region that has been implicated in motivational processes, such as seeking reward and avoiding harm. It contains dopamine (DA) neurons that project to limbic brain areas and give rise to the prominent mesolimbic DA pathway. In addition, it contains gamma-aminobutyric acid (GABA) neurons that also project to the limbic system as well as other major brain regions, such as the hindbrain and prefrontal cortex. Despite decades of research, the functions of VTA neurons remain mysterious and controversial. According to an influential hypothesis, VTA DA neurons encode a reward prediction error (RPE), a teaching signal that updates the value of learned associations (Schultz, Dayan, & Montague, 1997). It has also been proposed that VTA GABA neurons represent reward expectancy and provide the subtraction needed (actual reward minus expected reward), to compute a RPE (Eshel et al., 2015). According to another prominent hypothesis, however, VTA neurons encode the amount of effort, vigor, or ‘incentive’ we attribute to motivationally relevant stimuli (Salamone & Correa, 2012; Berridge & Robinson 1998). In most studies that attempt to relate VTA neural activity with reinforcement learning (RL) algorithms, the animals are often head-fixed and behavioral measures are usually limited to limb movements or licking. However, restraining the animal does not mean that they do not attempt to move their head and body. This creates a significant confound in all past research on DA neurons encoding RPE. I will argue that the conflict between the two prominent hypotheses of VTA function arises from both conceptual and empirical limitations, including the lack of precise and continuous behavioral measurements. To address these concerns, I first developed a novel head-fixation device that measures the forces exerted by the head in three orthogonal directions (up/down, left/right, forward/backward), as well as the forces exerted by the body (Chapter 2). The device contains load cells that convert analog voltage signals into continuous measures of force while the mice engage in traditional head-fixed tasks. By recording VTA neurons using in vivo electrophysiology and optogenetics while simultaneously measuring the continuous forces exerted by the mice, I found that VTA DA neurons encode the impulse vector (the magnitude and direction of force exerted over time) rather than RPE (Chapter 3). Moreover, according to the impulse-momentum theorem, I show how dynamic vector representations from head-fixed experiments can be translated into kinematic vector representations during freely moving behavior. According to the impulse-momentum theorem, impulse is equal to a change in momentum. In other words, a change in force is equivalent to a change in velocity assuming a constant mass, linking both dynamic and kinematic vector quantities. Then, by using the same continuous force measurements and manipulating the spatial location of reward during a traditional Pavlovian conditioning task, I falsified several key predictions from the RPE hypothesis (Chapter 4). By delivering the same reward in different locations (e.g., keeping the value and prediction constant), I was able to disambiguate an RPE signal from force exertion. I found that VTA DA neurons more precisely represented the impulse vector and not an RPE. Moreover, using a leaky integrator model, single unit activity of DA neurons could be used to predict the forces exerted across time regardless of reward predictability, as well as across multiple timescales. Then, I demonstrated that optogenetic manipulation of phasic DA activity has no impact on learning but directly modulates performance. At the same time, by using the same manipulations, I falsified the expectancy hypothesis of VTA GABA neurons and demonstrated they also represent vector quantities of force rather than expectancy. I found that VTA GABA neurons show opponent activity (increases or decreases of their firing rate) based on the direction of movement, despite the same level of expectancy and value (Chapter 4). Moreover, I utilized the leaky-integrator model to show that VTA GABA neurons represent the integral of DA activity. Finally, using in vivo electrophysiology, optogenetics, in vivo calcium imaging, and 3D motion capture during freely moving behavior in a novel reward tracking task, I found that a subset of VTA GABA neurons precisely represent three-dimensional rotational kinematics (Chapter 5). Taken together, these results demonstrate that the VTA controls the kinematics and dynamics necessary to control all motivated behaviors such as orientation, approach, and avoidance; whether to seek reward or avoid harm. These data unite the directional and activational components of motivation and provide precise physical quantities to influential concepts such as effort and vigor. Furthermore, I show the computational interaction between VTA DA and GABA neurons and demonstrate how they both participate in controlling the force vectors. Consequently, I made significant steps towards understanding how the VTA controls motivated behaviors and also falsified several key predictions of the RPE hypothesis, as well as improved the effort-related hypotheses. Thus, I have developed a new and comprehensive framework of VTA functioning.
This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 3.0 United States License.
Rights for Collection: Duke Dissertations
Works are deposited here by their authors, and represent their research and opinions, not that of Duke University. Some materials and descriptions may include offensive content. More info