Putting perception into action with inverse optimal control for continuous psychophysics

Continuous psychophysics: past, present, future

Continuous target-tracking psychophysics is an innovative experimental paradigm that has emerged as a powerful tool for studying perception, cognition, and visually guided behavior. This review outlines how continuous psychophysics complements traditional forced-choice methods by facilitating rapid data collection, providing insights into the real-time dynamics of perception and action, and enabling studies with special subject populations such as infants and patients. With its efficiency, conceptual simplicity, and ability to reveal temporal signatures of processing and performance, continuous psychophysics is poised to drive important advances across perception and cognitive science.

Temporal dynamics of human color processing measured using a continuous tracking task

We characterized the temporal dynamics of color processing using a continuous tracking paradigm by estimating subjects' temporal lag in tracking chromatic Gabor targets. To estimate the lag, we computed the cross-correlation between the velocities of the Gabor target's random walk and the velocities of the subject's tracking. Lag was taken as the time of the peak of the resulting cross-correlogram. We measured how the lag changes as a function of chromatic direction and contrast for stimuli in the LS cone contrast plane. In the same set of subjects, we also measured detection thresholds for stimuli with matched spatial, temporal, and chromatic properties. We created a model of tracking and detection performance to test whether a common representation of chromatic contrast accounts for both measures. The model summarizes the effect of chromatic contrast over different chromatic directions through elliptical isoperformance contours, the shapes of which are contrast independent. The fitted elliptical isoperformance contours have essentially the same orientation in the detection and tracking tasks. For the tracking task, however, there is a striking reduction in relative sensitivity to signals originating in the S cones.

A comparative analysis of perceptual noise in lateral and depth motion: Evidence from eye tracking

The characterization of how precisely we perceive visual speed has traditionally relied on psychophysical judgments in discrimination tasks. Such tasks are often considered laborious and susceptible to biases, particularly without the involvement of highly trained participants. Additionally, thresholds for motion-in-depth perception are frequently reported as higher compared to lateral motion, a discrepancy that contrasts with everyday visuomotor tasks. In this research, we rely on a smooth pursuit model, based on a Kalman filter, to quantify speed observational uncertainties. This model allows us to distinguish between additive and multiplicative noise across three conditions of motion dynamics within a virtual reality setting: random walk, linear motion, and nonlinear motion, incorporating both lateral and depth motion components. We aim to assess tracking performance and perceptual uncertainties for lateral versus motion-in-depth. In alignment with prior research, our results indicate diminished performance for depth motion in the random walk condition, characterized by unpredictable positioning. However, when velocity information is available and facilitates predictions of future positions, perceptual uncertainties become more consistent between lateral and in-depth motion. This consistency is particularly noticeable within ranges where retinal speeds overlap between these two dimensions. Significantly, additive noise emerges as the primary source of uncertainty, largely exceeding multiplicative noise. This predominance of additive noise is consistent with computational accounts of visual motion. Our study challenges earlier beliefs of marked differences in processing lateral versus in-depth motions, suggesting similar levels of perceptual uncertainty and underscoring the significant role of additive noise.

Precision and temporal dynamics in heading perception assessed by continuous psychophysics

It is a well-established finding that more informative optic flow (e.g., faster, denser, or presented over a larger portion of the visual field) yields decreased variability in heading judgements. Current models of heading perception further predict faster processing under such circumstances, which has, however, not been supported empirically so far. In this study, we validate a novel continuous psychophysics paradigm by replicating the effect of the speed and density of optic flow on variability in performance, and we investigate how these manipulations affect the temporal dynamics. To this end, we tested 30 participants in a continuous psychophysics paradigm administered in Virtual Reality. We immersed them in a simple virtual environment where they experienced four 90-second blocks of optic flow where their linear heading direction (no simulated rotation) at any given moment was determined by a random walk. We asked them to continuously indicate with a joystick the direction in which they perceived themselves to be moving. In each of the four blocks they experienced a different combination of simulated self-motion speeds (SLOW and FAST) and density of optic flow (SPARSE and DENSE). Using a Cross-Correlogram Analysis, we determined that participants reacted faster and displayed lower variability in their performance in the FAST and DENSE conditions than in the SLOW and SPARSE conditions, respectively. Using a Kalman Filter-based analysis approach, we found a similar pattern, where the fitted perceptual noise parameters were higher for SLOW and SPARSE. While replicating previous results on variability, we show that more informative optic flow can speed up heading judgements, while at the same time validating a continuous psychophysics as an efficient method for studying heading perception.

Context-dependence of deterministic and nondeterministic contributions to closed-loop steering control

In natural circumstances, sensory systems operate in a closed loop with motor output, whereby actions shape subsequent sensory experiences. A prime example of this is the sensorimotor processing required to align one's direction of travel, or heading, with one's goal, a behavior we refer to as steering. In steering, motor outputs work to eliminate errors between the direction of heading and the goal, modifying subsequent errors in the process. The closed-loop nature of the behavior makes it challenging to determine how deterministic and nondeterministic processes contribute to behavior. We overcome this by applying a nonparametric, linear kernel-based analysis to behavioral data of monkeys steering through a virtual environment in two experimental contexts. In a given context, the results were consistent with previous work that described the transformation as a second-order linear system. Classically, the parameters of such second-order models are associated with physical properties of the limb such as viscosity and stiffness that are commonly assumed to be approximately constant. By contrast, we found that the fit kernels differed strongly across tasks in these and other parameters, suggesting context-dependent changes in neural and biomechanical processes. We additionally fit residuals to a simple noise model and found that the form of the noise was highly conserved across both contexts and animals. Strikingly, the fitted noise also closely matched that found previously in a human steering task. Altogether, this work presents a kernel-based analysis that characterizes the context-dependence of deterministic and non-deterministic components of a closed-loop sensorimotor task.

Human navigation strategies and their errors result from dynamic interactions of spatial uncertainties

Goal-directed navigation requires continuously integrating uncertain self-motion and landmark cues into an internal sense of location and direction, concurrently planning future paths, and sequentially executing motor actions. Here, we provide a unified account of these processes with a computational model of probabilistic path planning in the framework of optimal feedback control under uncertainty. This model gives rise to diverse human navigational strategies previously believed to be distinct behaviors and predicts quantitatively both the errors and the variability of navigation across numerous experiments. This furthermore explains how sequential egocentric landmark observations form an uncertain allocentric cognitive map, how this internal map is used both in route planning and during execution of movements, and reconciles seemingly contradictory results about cue-integration behavior in navigation. Taken together, the present work provides a parsimonious explanation of how patterns of human goal-directed navigation behavior arise from the continuous and dynamic interactions of spatial uncertainties in perception, cognition, and action.

Movement trajectories as a window into the dynamics of emerging neural representations

The rapid transformation of sensory inputs into meaningful neural representations is critical to adaptive human behaviour. While non-invasive neuroimaging methods are the de-facto method for investigating neural representations, they remain expensive, not widely available, time-consuming, and restrictive. Here we show that movement trajectories can be used to measure emerging neural representations with fine temporal resolution. By combining online computer mouse-tracking and publicly available neuroimaging data via representational similarity analysis (RSA), we show that movement trajectories track the unfolding of stimulus- and category-wise neural representations along key dimensions of the human visual system. We demonstrate that time-resolved representational structures derived from movement trajectories overlap with those derived from M/EEG (albeit delayed) and those derived from fMRI in functionally-relevant brain areas. Our findings highlight the richness of movement trajectories and the power of the RSA framework to reveal and compare their information content, opening new avenues to better understand human perception.

Continuous psychophysics shows millisecond-scale visual processing delays are faithfully preserved in movement dynamics

Image differences between the eyes can cause interocular discrepancies in the speed of visual processing. Millisecond-scale differences in visual processing speed can cause dramatic misperceptions of the depth and three-dimensional direction of moving objects. Here, we develop a monocular and binocular continuous target-tracking psychophysics paradigm that can quantify such tiny differences in visual processing speed. Human observers continuously tracked a target undergoing Brownian motion with a range of luminance levels in each eye. Suitable analyses recover the time course of the visuomotor response in each condition, the dependence of visual processing speed on luminance level, and the temporal evolution of processing differences between the eyes. Importantly, using a direct within-observer comparison, we show that continuous target-tracking and traditional forced-choice psychophysical methods provide estimates of interocular delays that agree on average to within a fraction of a millisecond. Thus, visual processing delays are preserved in the movement dynamics of the hand. Finally, we show analytically, and partially confirm experimentally, that differences between the temporal impulse response functions in the two eyes predict how lateral target motion causes misperceptions of motion in depth and associated tracking responses. Because continuous target tracking can accurately recover millisecond-scale differences in visual processing speed and has multiple advantages over traditional psychophysics, it should facilitate the study of temporal processing in the future.

Modelling dataset bias in machine-learned theories of economic decision-making

Normative and descriptive models have long vied to explain and predict human risky choices, such as those between goods or gambles. A recent study reported the discovery of a new, more accurate model of human decision-making by training neural networks on a new online large-scale dataset, choices13k. Here we systematically analyse the relationships between several models and datasets using machine-learning methods and find evidence for dataset bias. Because participants' choices in stochastically dominated gambles were consistently skewed towards equipreference in the choices13k dataset, we hypothesized that this reflected increased decision noise. Indeed, a probabilistic generative model adding structured decision noise to a neural network trained on data from a laboratory study transferred best, that is, outperformed all models apart from those trained on choices13k. We conclude that a careful combination of theory and data analysis is still required to understand the complex interactions of machine-learning models and data of human risky choices.

Investigating cross-orientation inhibition with continuous tracking

We investigated cross-orientation inhibition with the recently developed continuous tracking technique. We designed an experiment where participants tracked the horizontal motion of a narrow vertical grating. The target was superimposed on one of three different backgrounds, in separate sessions: a uniform gray background or a sinusoidal grating oriented either parallel or orthogonal to the target. Both mask and target where phase reversed. We cross-correlated target and mouse movements and compared the peaks and lags of response with the different masks. Our results are in agreement with previous findings on cross-orientation inhibition: The orthogonal mask had a weak effect on the peaks and lags of correlation as a function of target contrast, consistently with a divisive effect of the mask, while the parallel mask acted subtractively on the response. Interestingly, lags of correlation decreased approximately linearly with contrast, with decrements of the order of 100 ms, even at 10 times the detection threshold, confirming that it is possible to investigate behavioral differences above threshold using the continuous tracking paradigm.

Out of control: computational dynamic control dysfunction in stress- and anxiety-related disorders

Control theory, which has played a central role in technological progress over the last 150 years, has also yielded critical insights into biology and neuroscience. Recently, there has been a surging interest in integrating control theory with computational psychiatry. Here, we review the state of the field of using control theory approaches in computational psychiatry and show that recent research has mapped a neural control circuit consisting of frontal cortex, parietal cortex, and the cerebellum. This basic feedback control circuit is modulated by estimates of reward and cost via the basal ganglia as well as by arousal states coordinated by the insula, dorsal anterior cingulate cortex, amygdala, and locus coeruleus. One major approach within the broader field of control theory, known as proportion-integral-derivative (PID) control, has shown promise as a model of human behavior which enables precise and reliable estimates of underlying control parameters at the individual level. These control parameters correlate with self-reported fear and with both structural and functional variation in affect-related brain regions. This suggests that dysfunctional engagement of stress and arousal systems may suboptimally modulate parameters of domain-general goal-directed control algorithms, impairing performance in complex tasks involving movement, cognition, and affect. Future directions include clarifying the causal role of control deficits in stress- and anxiety-related disorders and developing clinically useful tools based on insights from control theory.

Models of vision need some action

Bowers et al. focus their criticisms on research that compares behavioral and brain data from the ventral stream with a class of deep neural networks for object recognition. While they are right to identify issues with current benchmarking research programs, they overlook a much more fundamental limitation of this literature: Disregarding the importance of action and interaction for perception.

Neural networks need real-world behavior

Bowers et al. propose to use controlled behavioral experiments when evaluating deep neural networks as models of biological vision. We agree with the sentiment and draw parallels to the notion that "neuroscience needs behavior." As a promising path forward, we suggest complementing image recognition tasks with increasingly realistic and well-controlled task environments that engage real-world object recognition behavior.

Continuous psychophysics for two-variable experiments; A new "Bayesian participant" approach

Interest in continuous psychophysical approaches as a means of collecting data quickly under natural conditions is growing. Such approaches require stimuli to be changed randomly on a continuous basis so that participants can not guess future stimulus states. Participants are generally tasked with responding continuously using a continuum of response options. These features introduce variability in the data that is not present in traditional trial-based experiments. Given the unique weaknesses and strengths of continuous psychophysical approaches, we propose that they are well suited to quickly mapping out relationships between above-threshold stimulus variables such as the perceived direction of a moving target as a function of the direction of the background against which the target is moving. We show that modelling the participant in such a two-variable experiment using a novel "Bayesian Participant" model facilitates the conversion of the noisy continuous data into a less-noisy form that resembles data from an equivalent trial-based experiment. We also show that adaptation can result from longer-than-usual stimulus exposure times during continuous experiments, even to features that the participant is not aware of. Methods for mitigating the effects of adaptation are discussed.

Causal inference during closed-loop navigation: parsing of self- and object-motion

A key computation in building adaptive internal models of the external world is to ascribe sensory signals to their likely cause(s), a process of causal inference (CI). CI is well studied within the framework of two-alternative forced-choice tasks, but less well understood within the cadre of naturalistic action-perception loops. Here, we examine the process of disambiguating retinal motion caused by self- and/or object-motion during closed-loop navigation. First, we derive a normative account specifying how observers ought to intercept hidden and moving targets given their belief about (i) whether retinal motion was caused by the target moving, and (ii) if so, with what velocity. Next, in line with the modelling results, we show that humans report targets as stationary and steer towards their initial rather than final position more often when they are themselves moving, suggesting a putative misattribution of object-motion to the self. Further, we predict that observers should misattribute retinal motion more often: (i) during passive rather than active self-motion (given the lack of an efference copy informing self-motion estimates in the former), and (ii) when targets are presented eccentrically rather than centrally (given that lateral self-motion flow vectors are larger at eccentric locations during forward self-motion). Results support both of these predictions. Lastly, analysis of eye movements show that, while initial saccades toward targets were largely accurate regardless of the self-motion condition, subsequent gaze pursuit was modulated by target velocity during object-only motion, but not during concurrent object- and self-motion. These results demonstrate CI within action-perception loops, and suggest a protracted temporal unfolding of the computations characterizing CI. This article is part of the theme issue 'Decision and control processes in multisensory perception'.

Tasks and their role in visual neuroscience

Vision is widely used as a model system to gain insights into how sensory inputs are processed and interpreted by the brain. Historically, careful quantification and control of visual stimuli have served as the backbone of visual neuroscience. There has been less emphasis, however, on how an observer's task influences the processing of sensory inputs. Motivated by diverse observations of task-dependent activity in the visual system, we propose a framework for thinking about tasks, their role in sensory processing, and how we might formally incorporate tasks into our models of vision.