Understanding Perceptual Decisions by Studying Development and Neurodiversity

Abstract

A cornerstone of human information processing is how we make decisions about incoming sensory percepts. Much of psychological science has focused on understanding how these judgments operate in skilled adult observers. Although not typically the focus of this research, variability in how adults make these judgments is considerable. Here, we review complementary computational-modeling, electrophysiological-data, eye-tracking, and longitudinal approaches to the study of perceptual decisions across neurotypical development and in neurodivergent individuals. These data highlight multiple parameters and temporal dynamics feeding into how we become skilled adult perceptual decision makers, and they may help explain why we vary so much in how we make perceptual decisions.

Keywords

perceptual decisions neurodiversity temporal dynamics complementary methods

As skilled adult observers, we readily make accurate and fast decisions about sensory inputs of many different kinds, from dynamically changing moving stimuli to highly complex and cluttered visual scenes. Parameters underpinning optimal adult perceptual decision-making have been the focus of much research in psychological science. A perhaps less well understood question is how we come to differ from each other in making such decisions. Here, we outline evidence from complementary approaches in the developmental cognitive neurosciences that point to the need to understand differences at multiple time points of divergence in the complex dynamics leading to optimal perceptual decision-making. Computational modeling, temporally sensitive neuroimaging methods, and eye tracking provide rich data on how diversity in perceptual decision-making emerges for both neurotypical and neurodivergent individuals. We begin with evidence that when studies compare children with adults, parameters feeding into perceptual decisions differ across age groups from the onset of stimulus processing, taking moving stimuli as a central example. We then move onto developmental differentiation in the period during which we prepare to make perceptual decisions on the basis of our goals, prior knowledge, or information recently committed to memory. We conclude by providing an overview of evidence from autism, dyslexia, and genetic syndromes, suggesting multiple influences and routes to perceptual decision-making dynamics both before and after the onset of incoming stimuli.

The Developmental Dynamics of Perceptual Decisions: The Case of Motion Processing

Perceptual decisions have commonly been studied in adults using random-dot motion tasks. These tasks allow the strength of sensory evidence to be manipulated precisely, and the neurophysiological correlates of sensory encoding and decisional processes have been disentangled in nonhuman animals or in human adults performing these tasks (Kelly & O’Connell, 2015), making them useful for deconstructing multiple processes contributing to decisions. However, children provide a very interesting and far less explored model in which to study the emergence of decision parameters because they show a protracted development of sensitivity to motion information, with adult-like levels of sensitivity not being reached until mid to late childhood (Hadad et al., 2011; Manning et al., 2014). So far, most developmental studies have focused on characterizing the underlying sensory parameters affecting performance (Hadad et al., 2015; Meier & Giaschi, 2014) rather than considering developmental changes in the multiple cognitive processes underlying perceptual decisions. Yet by applying complementary methods such as modeling and electroencephalography, as well as a developmental perspective, we can uncover multiple processes that contribute to efficient adult perceptual decisions.

For example, Manning et al. (2014) used a computational-modeling technique to investigate contributors to age-related differences in motion-processing ability. They studied whether the development of motion coherence sensitivity is limited by internal noise (i.e., imprecision in estimating the directions of individual elements) and/or global pooling across local estimates by presenting moving dots under equivalent noise-direction discrimination conditions and in motion coherence conditions at both slow (1.5° per second) and fast (6° per second) speeds to children ages 5, 7, 9, and 11 years old and adults. As children got older, their levels of internal noise were reduced, and they were able to average across more local motion estimates, but age-related improvements in coherent motion perception were predicted by improvements in averaging and not by reductions in internal noise. In other words, improved performance in older children and adults was driven by how efficiently they combined motion signals over space rather than how precisely they estimated the direction of individual moving elements in a visual scene.

Additional insights into the dynamic evolution of processes contributing to perceptual decisions can be gained from electroencephalography (EEG) because of its high temporal resolution. We adapted a traditional motion coherence task for children and optimized it for EEG. One hundred white stimulus dots (described to the children as fireflies) were randomly positioned within a central square region on a black screen and moved at a speed of 6° per second. The dots had a limited lifetime of 200 ms (with randomized starting lifetimes), and dots moving outside the square stimulus region were wrapped around to the opposite side. Each trial consisted of a fixation period, a random-motion period, a stimulus period, and an offset period. The fixation period, during which only the central fixation square was shown, was presented for a randomly selected duration between 800 and 1,000 ms. The stimulus dots first appeared in the random-motion period, during which they moved in random, incoherent directions for a randomly selected duration between 800 and 1,000 ms. In the stimulus period, a proportion of the dots moved coherently either upward or downward (described as bees to be caught by a beekeeper), whereas the remainder of the dots continued to move in random directions. The stimulus period lasted until a response was made or until 2,500 ms had passed. Finally, an offset period continued the coherent stimulus presentation for a randomly selected duration between 200 and 400 ms. The jittered durations of the fixation, random-motion, and offset periods were intended to minimize expectancy effects.

While children and adults performed this motion coherence task, we collected high-density EEG data from all participants, and we found that age-related differences in motion sensitivity are accompanied by differences in both a late, sustained decision-related component and an earlier component reflecting sensory encoding (as shown in Fig. 1, adapted from Manning et al., 2019). The late decision-related component is separable and distinguishable from upstream sensory signals. It predicted response time even when the strength of evidence was held constant, and it scaled with motion coherence and increased as a function of time (criteria set out by Kelly & O’Connell, 2013). The other component did not show the association with reaction time (Manning et al., 2019). In addition, as we will discuss in greater detail later, when we used an EEG processing approach geared to deconvolving stimulus-locked and response-locked signals (Manning, Hassall, Hunt, Norcia, Wagenmakers, Evans, & Scerif, 2022; Manning, Hassall, Hunt, Norcia, Wagenmakers, Snowling, & Evans, 2022), the late decision-related component in the deconvolved EEG signal correlated with drift rate, suggesting that it quite specifically indexes evidence accumulation rather than, for example, boundary separation or sensorimotor processes. These results suggest that there are multiple and separable signatures of the dynamics at which perceptual decisions take shape and change from childhood to adulthood.

Fig. 1.

Age-related differences in electroencephalography (EEG) activity during perceptual decision-making tasks. Manning et al. (2019) extracted two patterns of EEG activity (“components”) when adults were making perceptual decisions about the overall direction of moving dots: a sustained component with maximal activity over centroparietal electrodes (over the middle of the head) that gradually increased over time (upper left image) and an earlier component with maximal activity over occipital electrodes (at the back of the head; lower left image). Following previous work, Manning et al. linked these components to the decision-making process and early sensory encoding of motion information, respectively. As shown in the graphs, the activity within both components showed age-related differences in amplitude and shape, with horizontal gray bars highlighting areas of significant differences between age groups. These results suggest that age-related changes in perceptual decisions are accompanied both by changes in early processes linked to sensory encoding and by later decision-making processes. Lighter color error bands indicate confidence intervals. Adapted from Manning et al. (2019).

Complementing these results, we highlighted how multiple parameters differentiate younger children from older children and adults using diffusion modeling (Manning et al., 2021)—a cognitive model that breaks accuracy and response time into distinct processing stages (Ratcliff & McKoon, 2008; Fig. 2). Specifically, younger children had lower drift rates (reduced sensitivity), wider boundary separation (increased response caution), and longer nondecision times than older children and adults. Moreover, the slope of an EEG component that was maximal over centroparietal electrodes was steeper in adults compared with younger children (Manning et al., 2019), and it correlated positively with drift rate (Manning et al., 2021), suggesting that evidence accumulation changes over age are accompanied by changes in a response-locked EEG component.

Fig. 2.

Schematic representation of the decision-making process in the diffusion model for a task that involves deciding between upward and downward motion. In the diffusion model, the decision-making process is represented as a noisy accumulation of evidence from a starting point toward one of two decision bounds. When a participant decides between upward and downward motion, the decision bounds correspond to “up” and “down” responses. In this representation, the overall motion in the stimulus is going upward. Boundary separation represents the width between the two bounds and reflects how cautious an observer is. Wider decision boundaries reflect that more evidence is required before making a decision (i.e., more cautious responses). Drift rate reflects the rate of evidence accumulation, which depends on both the individual’s sensitivity to a stimulus and the stimulus strength. Nondecision time is the time taken for sensory encoding processes prior to the decision-making process and response generation processes after a bound is reached. Manning, Hassall, Hunt, Norcia, Wagenmakers, Evans, and Scerif (2022) showed that younger children accumulate evidence more gradually (i.e., have a lower drift rate) and make decisions more cautiously (i.e., have wider decision bounds) than older children and adults, and they also take longer for nondecision processes (i.e., longer nondecision time). Adapted from Manning, Hassall, Hunt, Norcia, Wagenmakers, Snowling, and Evans (2022).

By taking a developmental perspective that leverages similarities and differences between children and adults, the complementary approaches discussed above have significantly advanced our ability to unveil the multiple processes building to perceptual decisions from stimulus onset to response generation. These approaches can extend even earlier in the timeline of processing to investigate the dynamics that precede incoming sensory stimuli and prepare an observer to make perceptual decisions.

Preparing to Make Perceptual Decisions: The Influence of Prior History

In addition to the temporal dynamics and multiple parameters building to the decision point from stimulus onset, the decisions we make are also influenced by how we prepare to make them. This is because of attentional biases from information we have encountered and committed to memory. This is the case when, for example, we have learned over time that stimuli for which we are preparing to make perceptual decisions occur at particular locations in complex visual scenes, whether static or dynamic. Attentional biases that depend on information recently encoded into memory (henceforth referred to as memory-guided attention) influence incoming percepts so that our prior history guides perceptual decision-making (Nobre & Stokes, 2019). Again, a multimethod developmental cognitive neuroscience approach highlights multiple factors contributing to how we prepare to make perceptual decisions.

Doherty et al. (2017) combined computational approaches, eye-tracking methodology, and individual-differences measures by asking participants to search for perceptual targets in complex visual scenes containing social or nonsocial distractors and repeating exposure to those scenes and targets over blocks to induce a learning history for their association. Eye tracking revealed significantly more attentional capture to the social compared with the nonsocial distractors embedded in each scene. Critically, participants’ memory precision for target locations was poorer for social scenes compared with nonsocial ones. The difference between social and nonsocial scenes was moderated by individual differences in social anxiety, with anxious individuals remembering target locations better under conditions of social distraction than nonanxious individuals, suggesting an interplay among attentional dynamics during learning about the scenes, memory for those scenes, and the characteristics of diverse observers, a point to which we will return in the Neurodiversity section below. Complementing this paradigm with EEG data collection, Doherty, van Ede, et al. (2019) showed that while they were preparing to make decisions about targets that had appeared in the context of social scenes, participants’ brain signatures of preparation differed for social and nonsocial targets. A lateralized modulation of alpha-band oscillations over one hemisphere compared with the other, a signature of preparatory EEG that is characteristic of attention while one is preparing to make perceptual decisions, was different when participants prepared to perceive targets that had been encoded into memory in the context of social scenes compared with nonsocial scenes. These findings highlighted that preparing to make perceptual decisions differed depending on the context in which targets had appeared, suggesting a dynamic interplay among memory-guided attention, the nature of visual scenes, and perceptual decisions about targets embedded in those scenes.

Investigating memory-guided attention via a developmental science perspective has further nuanced our understanding of adult perceptual decisions. Doherty, Fraser, et al. (2019) asked 6- to 10-year-old children and young adults to search for targets in complex visual scenes (following Doherty et al., 2017) while eye-tracking and simple manual responses were collected. Social stimuli distracted both children and adults during visual search, and eye tracking revealed even greater attentional capture by social distractors for children compared with adults. Intriguingly, however, children demonstrated overall better memory precision for target locations than adults. Children spent more time inspecting the whole scene before reporting target locations during the learning phase. We hypothesize that this led to greater exploration of and encoding of the full scene during the longer time inspecting it. This slower performance would normally be considered less efficient search performance, but this resulted in more precise later memory, suggesting retrieval of more precise target locations either via accessing scene gist and/or landmarks within it. A further replication, combining memory precision data with eye-tracking data collected during both the learning and memory phases would test this proposed mechanism more directly. Finally, when participants detected targets within visual scenes, adults were slower for targets appearing at unexpected locations within social scenes compared with nonsocial scenes, but this was not the case for children, again perhaps pointing to better learning of the scene as a whole. In contrast, fast-processing adults focused on locating targets are very likely to have suffered from more shallow processing of the highly salient and attention-grabbing social distractors. In other words, temporal search patterns that are slow and exploratory for children, rather than fast and goal driven for adults, may reduce interference from salient distractors during the attention-orienting phase. Using a similar memory-guided attention paradigm, Nussenbaum et al. (2019) also found evidence that children are better able than adults to draw from information encoded into memory when preparing to detect perceptual targets in complex visual scenes. Following on this work on memory-guided attention development, Shalev et al. (2022) recently found that children as young as 4 years old are as sensitive as adults to guiding attention to spatiotemporal regularities in dynamically changing scenes.

As a whole, these complementary data suggest that the interplay among attentional biases, memory, and memory-guided attention is complex, changing over developmental time, and moderated by individual differences across observers. In the next section, we will turn to precisely how studying neurodiversity can inform our understanding of perceptual decisions and their dynamics.

Neurodiversity

When one studies neurodiversity, the complex cascade of processes feeding into perceptual decisions is even more apparent. This is because differences at multiple stages of processing can converge or distinguish individuals with diverse profiles of development. Autism, which affects social communicative and nonsocial behaviors, and dyslexia, which affects reading ability, are distinct conditions that have both been linked to atypical motion sensitivity (Benassi et al., 2010; Van der Hallen et al., 2019). Multimethod approaches can shed light on where differences lie in autistic and dyslexic observers compared with neurotypical observers and identify areas of overlap and divergence between the conditions. Toffoli et al. (2021) investigated the temporal dynamics of EEG signals locked to the onset of directional motion stimuli and found that autistic and dyslexic children did not differ significantly from typically developing children in early stages of processing associated with early sensory encoding (in a component like that shown in Fig. 1, lower row). Yet both autistic and dyslexic children differed from typically developing children in later stages of processing, which are thought to involve processes such as segregation of signal from noise and decision-making.

Despite these similar patterns in EEG activity for autistic and dyslexic children, diffusion-modeling studies combined with response-locked EEG indices have uncovered differences in processing between autistic and dyslexic children. Dyslexic children showed reduced accumulation of motion evidence compared with typically developing children across two motion tasks that differed in their requirements for segregating signal from noise (Manning, Hassall, Hunt, Norcia, Wagenmakers, Snowling, & Evans, 2022). These differences were related to reductions in a response-locked EEG measure reflecting the decision-making process. Yet a comparable study (Manning, Hassall, Hunt, Norcia, Wagenmakers, Evans, & Scerif, 2022) showed no clear differences between autistic children and typically developing children in diffusion model parameters or the response-locked EEG measure. Future work is therefore needed to establish the conditions under which autistic children make atypical perceptual decisions about motion stimuli and why this is the case. There is also some evidence that attention-deficit/hyperactivity disorder (ADHD) symptoms—in particular, hyperactivity symptoms—might relate to perceptual decision-making parameters in autistic participants (Manning, Hassall, Hunt, Norcia, Wagenmakers, Evans, & Scerif, 2022). Therefore, by applying multimethod approaches, we can identify both areas of overlap and areas of difference in how neurodivergent individuals make perceptual decisions.

Of note, further valuable insights can come from neurodivergent individuals for whom symptoms of autism and attention differences cluster together. Fragile X syndrome (FXS) is the most common genetically identified and inherited monogenic syndrome associated with a high prevalence of autistic and ADHD symptoms (Doherty & Scerif, 2017). By tracking autism and ADHD symptoms longitudinally over the course of 3 years across a diverse group of children and teenagers with FXS, Doherty et al. (2020) found that individual variation in ADHD symptoms predicted later individual variation in autism symptoms, bolstering the need to investigate more precisely how attention, activity levels, and autistic characteristics interact over developmental time. Indeed, Guy et al. (2020) complemented the longitudinal approach by Doherty et al. (2020) by using the experimental memory-guided attention paradigm described earlier and by measuring the temporal dynamics of processing over time using eye tracking in both neurotypical children and young people, as well as children and young people with FXS. The study revealed differences in the dynamics of attention to social stimuli for children and young people with FXS compared with neurotypical participants: Children and young people with FXS inspected the social elements of complex visual scenes to a lesser degree than neurotypical individuals when they were first presented with them. This avoidance pattern decreased with experience of the scenes, but it was coupled with poorer memory for these scenes, again pinpointing the interplay among memory-guided preparation, stimulus processing, and perceptual decisions.

In their entirety, complementary multimethod data about how neurodivergent children and young people make perceptual decisions converge in suggesting that there are multiple ways in which observers can differ over the timeline leading to perceptual decisions: from preparation to processing of stimuli and decision-making, as well as in the short-term prior history and longitudinal change. Of note, we began by describing how a decision-theoretic framework can help disentangle distinct contributors to perceptual choice and how these parameters manifest in temporally distinguishable differences between neurotypical and neurodivergent children (e.g., slower drift rate for some neurodivergent children). A parallel set of findings points again to dynamics, suggesting that preparatory and memory-dependent dynamics influence perceptual decisions differently across ages. However, the two literatures have not yet been bridged: That is, to our knowledge, developmental scientists have not yet tested whether prestimulus dynamics (preparation, memory-dependent attention effects) change poststimulus decisional parameters (e.g., drift rate, decision bounds) rather than simply affect perceptual choices. What is also not known is whether preparatory dynamics influence decisional parameters differently in neurotypical and neurodivergent children. Future research could integrate approaches to prestimulus and predecision dynamics for both neurodivergent and neurotypical development from childhood into adulthood.

Conclusion

We have reviewed evidence that multiple complex parameters contribute to perceptual decisions and that these parameters can differ in informative ways across development and with neurodiversity. A better understanding of the developmental processes involved may help us, in turn, to understand the perceptual decisions made by skilled adult observers and, in particular, why there is considerable variability in how people make perceptual decisions.

Footnotes

Acknowledgements

A CC BY or equivalent license is applied to the author accepted manuscript arising from this submission in accordance with the Wellcome Trust grant’s open-access conditions.

Transparency

Action Editor: Robert L. Goldstone

Editor: Robert L. Goldstone

ORCID iD

Gaia Scerif

References

Benassi

Simonelli

Giovagnoli

Bolzani

(2010). Coherence motion perception in developmental dyslexia: A meta-analysis of behavioral studies. Dyslexia, 16(4), 341–357.

Doherty

B. R.

Fraser

Nobre

A. C.

Scerif

(2019). The functional consequences of social attention on memory precision and on memory-guided orienting in development. Developmental Cognitive Neuroscience, 36, Article 100625. https://doi.org/10.1016/j.dcn.2019.100625

Doherty

B. R.

Longhi

Cole

Karmiloff-Smith

Cornish

Scerif

(2020). Disentangling autism spectrum and attention-deficit/hyperactivity symptoms over development in fragile X syndrome. Research in Developmental Disabilities, 104, Article 103692. https://doi.org/10.1016/j.ridd.2020.103692

Doherty

B. R.

Patai

E. Z.

Duta

Nobre

A. C.

Scerif

(2017). The functional consequences of social distraction: Attention and memory for complex scenes. Cognition, 158, 215–223. https://doi.org/10.1016/j.cognition.2016.10.015

Doherty

B. R.

Scerif

(2017). Genetic syndromes and developmental risk for autism spectrum and attention deficit hyperactivity disorders: Insights from fragile X syndrome. Child Development Perspectives, 11(3), 161–166.

Doherty

B. R.

van Ede

Fraser

Patai

E. Z.

Nobre

A. C.

Scerif

(2019). The functional consequences of social attention for memory-guided attention orienting and anticipatory neural dynamics. Journal of Cognitive Neuroscience, 31(5), 686–698. https://doi.org/10.1162/jocn_a_01379

Guy

Ng-Cordell

Doherty

B. R.

Duta

Scerif

(2020). Understanding attention, memory and social biases in fragile X syndrome: Going below the surface with a multi-method approach. Research in Developmental Disabilities, 104, Article 103693. https://doi.org/10.1016/j.ridd.2020.103693

Hadad

Schwartz

Maurer

Lewis

T. L.

(2015). Motion perception: A review of developmental changes and the role of early visual experience. Frontiers in Integrative Neuroscience, 9, Article 49. https://doi.org/10.3389/fnint.2015.00049

Hadad

B. S.

Maurer

Lewis

T. L.

(2011). Long trajectory for the development of sensitivity to global and biological motion. Developmental Science, 14(6), 1330–1339.

10.

Kelly

S. P.

O’Connell

R. G.

(2013). Internal and external influences on the rate of sensory evidence accumulation in the human brain. The Journal of Neuroscience: The Official Journal of the Society for Neuroscience, 33(50), 19434–19441. https://doi.org/10.1523/JNEUROSCI.3355-13.2013

11.

Kelly

S. P.

O’Connell

R. G.

(2015). The neural processes underlying perceptual decision making in humans: Recent progress and future directions. Journal of Physiology, 109(1–3), 27–37.

12.

Manning

Dakin

S. C.

Tibber

M. S.

Pellicano

(2014). Averaging, not internal noise, limits the development of coherent motion processing. Developmental Cognitive Neuroscience, 10, 44–56.

13.

Manning

Hassall

C. D.

Hunt

L. T.

Norcia

A. M.

Wagenmakers

E.-J.

Evans

N. J.

Scerif

(2022). Behavioural and neural indices of perceptual decision-making in autistic children during visual motion tasks. Scientific Reports, 12(1), Article 6072. https://doi.org/10.1038/s41598-022-09885-4

14.

Manning

Hassall

C. D.

Hunt

L. T.

Norcia

A. M.

Wagenmakers

E.-J.

Snowling

M. J.

Evans

N. J.

(2022). Visual motion and decision-making in dyslexia: Reduced accumulation of sensory evidence and related neural dynamics. Journal of Neuroscience, 42(1), 121–134.

15.

Manning

Kaneshiro

Kohler

P. J.

Duta

Scerif

Norcia

A. M.

(2019). Neural dynamics underlying coherent motion perception in children and adults. Developmental Cognitive Neuroscience, 38, Article 100670. https://doi.org/10.1016/j.dcn.2019.100670

16.

Manning

Wagenmakers

E.-J.

Norcia

A. M.

Scerif

Boehm

(2021). Perceptual decision-making in children: Age-related differences and EEG correlates. Computational Brain & Behavior, 4(1), 53–69.

17.

Meier

Giaschi

(2014). The maturation of global motion perception depends on the spatial and temporal offsets of the stimulus. Vision Research, 95, 61–67.

18.

Nobre

A. C.

Stokes

M. G.

(2019). Premembering experience: A hierarchy of time-scales for proactive attention. Neuron, 104(1), 132–146. https://doi.org/10.1016/j.neuron.2019.08.030

19.

Nussenbaum

Scerif

Nobre

A. C.

(2019). Differential effects of salient visual events on memory-guided attention in adults and children. Child Development, 90(4), 1369–1388. https://doi.org/10.1111/cdev.13149

20.

Ratcliff

McKoon

(2008). The diffusion decision model: Theory and data for two-choice decision tasks. Neural Computation, 20(4), 873–922.

21.

Shalev

Boettcher

Wilkinson

Scerif

Nobre

N. C.

(2022). Be there on time: Spatial-temporal regularities guide young children’s attention in dynamic environments. Child Development, 93(5), 1414–1426. https://doi.org/10.1111/cdev.13770

22.

Toffoli

Scerif

Snowling

M. J.

Norcia

A. M.

Manning

(2021). Global motion evoked potentials in autistic and dyslexic children: A cross-syndrome approach. Cortex, 143, 109–126.

23.

Van der Hallen

Manning

Evers

Wagemans

. (2019). Global motion perception in autism spectrum disorder: A meta-analysis. Journal of Autism and Developmental Disorders, 49(12), 4901–4918.

Understanding Perceptual Decisions by Studying Development and Neurodiversity

Abstract

Keywords

The Developmental Dynamics of Perceptual Decisions: The Case of Motion Processing

Preparing to Make Perceptual Decisions: The Influence of Prior History

Neurodiversity

Conclusion

Recommended Reading

Footnotes

Acknowledgements

Transparency

ORCID iD

References