Explore the vast range of titles available.

Prediction error signals enable us to learn through experience. These experiences include economic choices between different rewards that vary along multiple dimensions. Therefore, an ideal way to reinforce economic choice is to encode a prediction error that reflects the subjective value integrated across these reward dimensions. Previous studies demonstrated that dopamine prediction error responses reflect the value of singular reward attributes that include magnitude, probability, and delay. Obviously, preferences between rewards that vary along one dimension are completely determined by the manipulated variable. However, it is unknown whether dopamine prediction error responses reflect the subjective value integrated from different reward dimensions. Here, we measured the preferences between rewards that varied along multiple dimensions, and as such could not be ranked according to objective metrics. Monkeys chose between rewards that differed in amount, risk, and type. Because their choices were complete and transitive, the monkeys chose “as if” they integrated different rewards and attributes into a common scale of value. The prediction error responses of single dopamine neurons reflected the integrated subjective value inferred from the choices, rather than the singular reward attributes. Specifically, amount, risk, and reward type modulated dopamine responses exactly to the extent that they influenced economic choices, even when rewards were vastly different, such as liquid and food. This prediction error response could provide a direct updating signal for economic values.

Utility is the fundamental variable thought to underlie economic choices. In particular, utility functions are believed to reflect preferences toward risk, a key decision variable in many real-life situations. To assess the validity of utility representations, it is therefore important to examine risk preferences. In turn, this approach requires formal definitions of risk. A standard approach is to focus on the variance of reward distributions (variance-risk). In this study, we also examined a form of risk related to the skewness of reward distributions (skewness-risk). Thus, we tested the extent to which empirically derived utility functions predicted preferences for variance-risk and skewness-risk in macaques. The expected utilities calculated for various symmetrical and skewed gambles served to define formally the direction of stochastic dominance between gambles. In direct choices, the animals’ preferences followed both second-order (variance) and third-order (skewness) stochastic dominance. Specifically, for gambles with different variance but identical expected values (EVs), the monkeys preferred high-variance gambles at low EVs and low-variance gambles at high EVs; in gambles with different skewness but identical EVs and variances, the animals preferred positively over symmetrical and negatively skewed gambles in a strongly transitive fashion. Thus, the utility functions predicted the animals’ preferences for variance-risk and skewness-risk. Using these well-defined forms of risk, this study shows that monkeys’ choices conform to the internal reward valuations suggested by their utility functions. This result implies a representation of utility in monkeys that accounts for both variance-risk and skewness-risk preferences.

Economic theories posit reward probability as one of the factors defining reward value. Individuals learn the value of cues that predict probabilistic rewards from experienced reward frequencies. Building on the notion that responses of dopamine neurons increase with reward probability and expected value, we asked how dopamine neurons in monkeys acquire this value signal that may represent an economic decision variable. We found in a Pavlovian learning task that reward probability-dependent value signals arose from experienced reward frequencies. We then assessed neuronal response acquisition during choices among probabilistic rewards. Here, dopamine responses became sensitive to the value of both chosen and unchosen options. Both experiments showed also the novelty responses of dopamine neurones that decreased as learning advanced. These results show that dopamine neurons acquire predictive value signals from the frequency of experienced rewards. This flexible and fast signal reflects a specific decision variable and could update neuronal decision mechanisms. Learning to choose the most fulfilling reward from a number of options requires the ability to infer the value of each option from unpredictable and changing environments. Neurons in the brain that produce a chemical called dopamine are critical for this learning process. They broadcast a 'prediction error' signal that alerts other areas of the brain to the difference between the actual reward and the previously predicted reward. Previous studies have shown that when dopamine neurons signal a prediction error, new learning about the value of an option takes place. To understand exactly what happens during this learning process, Lak et al. recorded electrical activity from dopamine neurons in the brains of two monkeys. Over a number of trials, the monkeys were shown one of three different novel images, each of which was associated with a different likelihood of receiving a large amount of a fruit juice reward. The recordings showed that the dopamine response to cues was divided into early and late components. At the start of learning, when the monkeys were unfamiliar with the likelihood that each image would yield a large juice reward, the early part of the dopamine response was large. The size of this part of the response decreased as the monkeys became more familiar with each image. The later part of the dopamine response changed to reflect the rewards the monkeys had received on previous trials. On trials where a reward was delivered, this part of the response grew larger, but diminished if a reward was not given. When the monkeys had to choose between rewards, the dopamine response was larger when the monkey chose the higher valued option over the lesser valued one, and smaller when the opposite choice was made, thus reflecting the animal’s choice. These choice-dependent responses were also sensitive to the value of unchosen option, and therefore, reflected the difference between the value of chosen and unchosen options. Future studies are now required to find out how manipulating the activity of the dopamine neurons influences the way animals learn and make decisions.

Gene therapy is an experimental approach to treating disease that involves altering faulty genes or replacing them with new, working copies. Most often, the new genetic material is delivered into cells using a modified virus that no longer causes disease, called a viral vector. Virus-mediated gene therapies are currently being explored for degenerative eye diseases, such as retinitis pigmentosa, and neurological disorders, like Alzheimer’s and Parkinson’s disease. A number of gene therapies have also been approved for treating some rare cancers, blood disorders and a childhood form of motor neuron disease. Despite the promise of virus-mediated gene therapy, there are significant hurdles to its widespread success. Viral vectors need to deliver enough genetic material to the right cells without triggering an immune response or causing serious side effects. Selecting an optimal vector is key to achieving this. A type of viruses called adeno-associated viruses (AAV) are prime candidates, partly because they can be easily engineered. However, accurately comparing the safety and efficacy of newly engineered AAVs is difficult, due to variation between test subjects and the labor and cost involved in careful testing. Öztürk et al. addressed this issue by developing an experimental pipeline called scAAVengr for comparing gene therapy vectors head-to-head. The process involves tagging potential AAV vectors with unique genetic barcodes, which can then be detected and quantified in individual cells using a technique called single-cell RNA sequencing. This means that when several vectors are used to infect lab-grown cells or a test animal at the same time, they can be tracked. The vectors can then be ranked on their ability to infect specific cell types and deliver useful genetic material. Using scAAVengr, Öztürk et al. compared viral vectors designed to target the light-sensitive cells of the retina, which allow animals to see. First, a set of promising viral vectors were evaluated using the scAAVengr pipeline in the eyes of marmosets and macaques, two small primates. Precise levels and locations of gene delivery were quantified. The top-performing vector was then identified and used to deliver Cas9, a genome editing tool, to primate retinas. Öztürk et al. also used scAAVengr to compare viral vectors in mice, analysing the vectors’ ability to deliver their genetic cargo to the brain, heart, and liver. These experiments demonstrated that scAAVengr can be used to evaluate vectors in multiple tissues and in different organisms. In summary, this work outlines a method for identifying and precisely quantifying the performance of top-performing viral vectors for gene therapy. By aiding the selection of optimal viral vectors, the scAAVengr pipeline could help to improve the success of preclinical studies and early clinical trials testing gene therapies.

Economic deliberations are slow, effortful and intentional searches for solutions to difficult economic problems. Although such deliberations are critical for making sound decisions, the underlying reasoning strategies and neurobiological substrates remain poorly understood. Here two nonhuman primates performed a combinatorial optimization task to identify valuable subsets and satisfy predefined constraints. Their behavior revealed evidence of combinatorial reasoning—when low-complexity algorithms that consider items one at a time provided optimal solutions, the animals adopted low-complexity reasoning strategies. When greater computational resources were required, the animals approximated high-complexity algorithms that search for optimal combinations. The deliberation times reflected the demands created by computational complexity—high-complexity algorithms require more operations and, concomitantly, the animals deliberated for longer durations. Recurrent neural networks that mimicked low- and high-complexity algorithms also reflected the behavioral deliberation times and were used to reveal algorithm-specific computations that support economic deliberation. These findings reveal evidence for algorithm-based reasoning and establish a paradigm for studying the neurophysiological basis for sustained deliberation. The authors develop a reward optimization framework to study sustained deliberation in nonhuman primates. As the computational complexity increased, animals deliberated longer and applied more complex reasoning strategies to optimize rewards.

Recent discoveries of extreme cellular diversity in the brain warrant rapid development of technologies to access specific cell populations within heterogeneous tissue. Available approaches for engineering-targeted technologies for new neuron subtypes are low yield, involving intensive transgenic strain or virus screening. Here, we present Specific Nuclear-Anchored Independent Labeling (SNAIL), an improved virus-based strategy for cell labeling and nuclear isolation from heterogeneous tissue. SNAIL works by leveraging machine learning and other computational approaches to identify DNA sequence features that confer cell type-specific gene activation and then make a probe that drives an affinity purification-compatible reporter gene. As a proof of concept, we designed and validated two novel SNAIL probes that target parvalbumin-expressing (PV+) neurons. Nuclear isolation using SNAIL in wild-type mice is sufficient to capture characteristic open chromatin features of PV+ neurons in the cortex, striatum, and external globus pallidus. The SNAIL framework also has high utility for multispecies cell probe engineering; expression from a mouse PV+ SNAIL enhancer sequence was enriched in PV+ neurons of the macaque cortex. Expansion of this technology has broad applications in cell type-specific observation, manipulation, and therapeutics across species and disease models.

Muller et al. demonstrate that reward signals recorded from the frontal cortex of nonhuman primates exhibit a population-based scheme for learning probability distributions over reward values. This study provides evidence that neural signals outside of the midbrain reflect the principles of distributional reinforcement-learning theory.

Dopamine prediction error responses are essential components of universal learning mechanisms. However, it is unknown whether individual dopamine neurons reflect the shape of reward distributions. Here, we used symmetrical distributions with differently weighted tails to investigate how the frequency of rewards and reward prediction errors influence dopamine signals. Rare rewards amplified dopamine responses, even when conventional prediction errors were identical, indicating a mechanism for learning the complexities of real-world incentives. Rare rewards amplify dopamine neuron responses, even when conventionally defined prediction errors are identical. This suggests that individual dopamine neurons are sensitive to predicted reward distributions and can facilitate the learning of complex incentive structures.

Rewards are fundamental to everyday life. They confer pleasure, support learning, and mediate decisions. Dopamine-releasing neurons in the midbrain are critical for reward processing. These neurons receive input from more than 30 brain areas and send widespread projections to the basal ganglia and frontal cortex. Their phasic responses are tuned to rewards. Specifically, dopamine signals code reward prediction error, the difference between received and predicted rewards. Decades of research in awake, behaving non-human primates (NHP), have shown the importance of these neural signals for learning and decision making. In this review, we will provide an overview of the bedrock findings that support the reward prediction error hypothesis and examine evidence that this signal plays a role in learning and decision making. In addition, we will highlight some of the conceptual challenges in dopamine neurophysiology and identify future areas of research to address these challenges. Keeping with the theme of this special issue, we will focus on the role of NHP studies in understanding dopamine neurophysiology and make the argument that primate models are essential to this line of research.