Human Psychophysics: The Foundation
Psychophysics, the study of the relationship between physical stimuli and perceptual experiences, has a rich tradition in human research, dating back to pioneers like Gustav Fechner and Ernst Weber. By designing tasks that precisely control stimulus properties, experimenters have been able to quantify perceptual abilities and decision-making processes with remarkable rigor. For instance, using well-parameterized tasks, early psychophysicists were able to explore fundamental sensory thresholds—such as the minimum intensity of light detectable in a dark room or the smallest detectable difference in weight between two objects. By adjusting parameters like intensity, duration, or spatial frequency, these studies provided structured insights into how the brain processes sensory information and formed the basis of our knowledge on perceptual acuity and sensory integration.
As psychophysics advanced, so too did the sophistication of tasks and measurements used to probe more complex processes, such as motion coherence in a moving dot field or the contrast of Gabor patches in visual perception studies. Researchers began tracking behavioral metrics like choice probability, reaction time, and accuracy to map out the underlying computations of perception and judgment. The controlled nature of these tasks allowed for precise manipulation of sensory input, which made it possible to model the resulting behavioral outputs with mathematical precision. Furthermore, the responses collected from these experiments could be interpreted using normative modelssuch as drift diffusion models for perceptual decision-making. These models provide interpretive parameters that not only explain the observed behaviors but also offer a structured approach for linking behavior with brain activity. Psychophysics, thus, not only illuminated sensory processing in humans but laid the groundwork for exploring cognition and perception in ways that would later be adapted for studying neural circuits in non-human animals.
Extending to Higher Cognition
Building on the success of perceptual studies, researchers began applying the same rigor of well-parameterized tasks to explore more complex cognitive functions such as working memory, multi-step planning, and decision-making. These higher cognitive processes, which go beyond basic sensory perception, required innovative task designs that could isolate specific mental operations. For instance, tasks like the N-back are used to assess working memory by requiring participants to keep track of a sequence of stimuli over time, while multi-step planning tasks may present a maze or a problem that requires a series of strategic choices. These types of tasks demand more than just perception—they engage neural circuits involved in memory, attention, and executive function, providing insight into how the brain organizes and manipulates information.
The introduction of such tasks created what is often termed a “behavioral clamp”—a controlled setting in which specific cognitive processes are activated reliably, making it possible to measure and model them with precision. When paired with functional brain imaging techniques like fMRI or electrophysiological recordings such as EEG, researchers could observe not only behavioral outputs but also the brain regions and networks activated during these tasks. This combination of behavioral precision with neural measurements allowed for a more detailed understanding of how cognitive operations are represented in the brain. For example, the use of fMRI in conjunction with multi-step decision-making tasks directly exposes the concurrent activation of the prefrontal cortex and basal ganglia, linking executive function with reward processing. Similarly, EEG studies could track the real-time brain dynamics that accompany choices, memory recall, or error detection, mapping out the temporal flow of information processing.
Importantly, the data from these higher cognitive tasks can also be interpreted through normative models like reinforcement learning for decision-making or Bayesian inference models for probabilistic reasoning. These models help distill the cognitive processes underlying task performance into quantifiable parameters, which in turn can be related back to neural signals. For instance, reinforcement learning models provide parameters such as learning rate or exploration-exploitation trade-offs, which offer mechanistic insights into how the brain might update strategies or adapt to changing environments. By fitting these models to task performance data, researchers could infer the underlying cognitive strategies and relate them to specific brain regions or patterns of brain activity, enriching our understanding of the neural basis of human cognition.
Through this structured approach, the field has uncovered valuable insights into human cognition, establishing a framework for systematically probing complex brain functions that would later be adapted for non-human primate and rodent studies. These human experiments thus paved the way for studying brain computations not only in isolation but also as part of a broader cognitive architecture, setting the stage for cross-species exploration of brain function.
Non-Human Primates: Precision in Neural Coding
The success of well-parameterized tasks in human cognitive research was followed by their application to non-human primates, such as macaque monkeys, whose research aspires to model certain aspects of human mental processing. By training monkeys on tasks requiring visual fixation, visuospatial attention, and visual motion processing, people have gained insight into the neural correlates of perception and attention in a species capable of complex, human-like responses (in certain domains). Non-human primates are highly trainable and share many homologous brain structures with humans, making them invaluable for studies requiring both behavioral complexity and precise neural recordings. Tasks adapted for monkeys, such as the delayed match-to-sample for working memory or visual search tasks for attention, closely mirror the structured nature of human experiments but allow for a far more detailed look at the neural code underlying these processes.
One major advantage of non-human primate studies is the ability to record single-neuron activity in real-time. Unlike techniques used in human studies, which generally capture broad neural activity (e.g., fMRI measures blood flow changes across thousands of neurons), single-neuron recordings in monkeys enable pinpointing the activity of individual neurons and neuron populations involved in a task. This capability provides an incredibly detailed view of how specific neurons encode and compute sensory information, motor plans, and decision outcomes.
Furthermore, single-neuron recordings enable a unique approach to studying cognitive processes like attention, working memory, and decision-making. One can observe, for instance, how neurons in the prefrontal cortex change their firing patterns as attention is directed toward a stimulus or held in working memory over a delay period. These dynamic changes in neural activity provide clues about the coding mechanisms for cognitive control, goal-directed behavior, and information maintenance—processes that are more complex and nuanced in non-human primates than in rodents. In multi-step planning or decision-making tasks, researchers can track the sequence of neural firing patterns as monkeys evaluate their options and make choices, offering a granular look at how the brain integrates sensory information with learned rules and expected outcomes.
Another critical aspect of non-human primate research is the ability to decode neural signals directly and assess how they correspond to specific task parameters or behavioral choices. By linking single-neuron activity with task-based behaviors, researchers can validate and refine normative models derived from human studies, such as drift diffusion models or Bayesian inference models, at an incredibly fine level. This level of detail provides a complementary perspective to human studies, enabling a more comprehensive understanding of the neural code underlying cognitive processes and offering testable hypotheses for interpreting human neural data
Rodents: The Power of Causal Tools
While well-parameterized tasks have transformed our understanding of perception and cognition in humans and non-human primates, applying this rigor to rodents has unlocked an entirely new realm of possibilities, particularly in the domain of causal manipulation. Historically, rodent research focused on tasks with more degrees of freedom and little structure or parameter control. However, the introduction of well-parameterized tasks in rodents—such as two-alternative forced-choice (2AFC) for perceptual decision making—has enabled the application of structured, rigorous task design of human psychophysics to smaller animals, providing a pathway to study fundamental computations with unprecedented mechanistic resolution.
The unique strength of rodent research lies in combining controlled tasks, neural measurements and optogenetics, which enables the manipulation of specific neural populations with extraordinary speed and precision. Using light to activate or inhibit targeted neurons, optogenetics can directly probe neural circuits at the millisecond timescale, matching the speed of natural neural processing. This capability allows for investigating how specific neurons or circuits causally contribute to behaviors in real-time. For instance, in a visual discrimination task where rodents decide on the orientation of a Gabor patch, one can selectively stimulate or silence neurons in the visual cortex precisely when stimuli are presented. By observing how these manipulations affect performance, one can expose the causal link between specific patterns of neural activity and perceptual decision-making, elucidating the role of neural circuits in fundamental computations.
This type of rodent behavioral data can also be fit by normative model, and directly link model parameters (e.g., decision thresholds, learning rates) to neural activity. Causal tools can directly validate these fits and their relationship to neural activity patterns and/or behavioral outcomes.
While rodent models offer a unique level of control over neural circuits, there are inherent limitations when it comes to studying higher-level cognitive functions. Rodents, though capable of learning and performing complex tasks, do not possess the advanced working memory, abstract reasoning, or planning faculties observed in primates. This creates a ceiling to our understanding of higher cognition based solely on rodent studies, underscoring the value of a cross-species approach.
The importance of Cross-Species studies
With the rise of well-parameterized tasks and precision tools in neuroscience, causal manipulations are now an integral aspect of research across species, each adding unique insights into brain function and computation. While optogenetics and other causal techniques were initially developed for rodent models, they are increasingly being adapted for use in non-human primates, providing the potential to explore more sophisticated cognitive functions with causal precision. This cross-species approach leverages each model’s strengths: rodents for detailed circuit analysis and precise manipulation, non-human primates for higher-order cognitive tasks closer to human cognition, and humans for investigating uniquely human capabilities.
The move to optogenetics in non-human primates is enabling experimenters to link specific neural circuits to behaviors with unprecedented specificity in a species capable of complex tasks and social behaviors. However, while non-human primates are invaluable for studying higher-order cognition, unique aspects of human cognition, such as abstract reasoning, language processing, and complex social behavior, remain beyond their scope. In humans, single-neuron recording capabilities have recently become available, mostly through collaboration with clinical neurosurgery patients. These recordings provide a window into uniquely human cognitive abilities, allowing researchers to examine individual neuron responses during tasks involving complex reasoning or language comprehension. This is especially valuable because these capabilities are distinct to humans and require the highest-resolution insight into the neural code for human-specific cognitive functions.
Human single-neuron recording studies have led to new discoveries in areas such as episodic memory, social processing, and abstract reasoning. For instance, single-neuron recordings in the medial temporal lobe have demonstrated neuron populations that respond selectively to specific memory cues, effectively serving as neural markers for individual memories. Such findings highlight how human-specific neural coding mechanisms operate within broader cognitive architectures. However, human studies are naturally constrained by the limited contexts in which electrodes can be placed—typically restricted to clinical cases where electrode implantation is necessary for therapeutic reasons. Thus, while human recordings offer remarkable insights, they are limited to specific brain regions and contexts.
Causal manipulations using genetically-based tools like optogenetics are unlikely to be implemented in humans due to ethical and technical constraints. As such, rodent and non-human primate models remain crucial for probing the neural basis of cognition at a level of causality that is not feasible in human studies. For instance, in rodents, we can modulate neural activity in a single cell or a small group of cells within circuits that support memory, reward, or decision-making, and observe the direct effects on behavior. This level of precision is invaluable for testing hypotheses about circuit function and validating models of computation that might later be refined and examined in non-human primate and human contexts.
The development of miniaturized, wireless recording systems and single-cell optogenetics has further advanced cross-species research by allowing the study of natural behaviors and social interactions in freely moving animals. These innovations are proving invaluable for understanding neural computations within complex, naturalistic contexts that closely resemble the environments in which cognition naturally operates. Additionally, optogenetics in non-human primates is advancing our ability to perform causal manipulations in the study of high-level cognition, enabling researchers to make specific, time-locked neural adjustments while animals engage in cognitively demanding tasks.
This progression in causal tools across species represents a continuum in neuroscience, where each species adds a unique perspective. Rodents provide unparalleled causal precision for dissecting foundational computations, non-human primates allow us to study complex cognition within neural circuits homologous to those of humans, and humans bring us insights into uniquely human cognitive capacities. This cross-species toolkit, integrating well-parameterized tasks and causal manipulations, is rapidly advancing our understanding of the neural basis of cognition, perception, and behavior, establishing a comprehensive framework for the future of neuroscience.