Neuroscience and Theory in Child Psychology: Uncomfortable Bedfellows?

It has long been believed that new insights from cognitive neuroscience will require us to change our theories for understanding children’s cognitive development. Developmental theories, exemplified by Piaget’s theory of cognitive development (Piaget, 1954), are explanatory systems that give a causal account for the changes that we observe in the developing infant and child. When a theory offers a good explanatory system for what can be observed behaviourally, it can have far-reaching effects. Again, this is exemplified by Piaget’s theory, which has dominated child psychology for over half a century. In this special issue, four quite different perspectives of the implications of neuroscience for developmental theory and vice versa are offered. These perspectives are based on invited plenaries originally delivered to the Jean Piaget Society. Most of the evidence in the different papers focuses on developmental cognitive neuroscience. Developmental cognitive neuroscience studies offer the seductive possibility of revealing directly the nature of children’s mental representations and how they change in response to cultural or contextual factors, education, or maturation. To date, however, this apparent promise has not really been cashed out. Too much prior work has ignored the inherent limitations of current brain imaging methods. A good developmental theory should tell us why development takes the path that we observe. Almost all current cognitive neuroimaging studies focus on where in the brain something is happening and tell us about correlations between neural activity and behaviour.Piaget was a constructivist, and many of his original theoretical ideas have stood up remarkably well despite the advent of cognitive neuroscience (Goswami, 2019). He was a biologist by training, and in his theory the primary causal mechanism for building knowledge was the adaptation to and refinement of existing cognitive “schemes” by the environment. Knowledge was constructed by the child as a consequence of his or her active experiences with the external world. In sympathy with a constructivist approach, the paper by Arsiladou offers a descriptive level analysis of how neo-Piagetian theories may be amenable to input from neuroscience. Focusing on the theory of constructive operators pioneered by Pascaul-Leone (1970), Arsiladou links different operators in that theory (Pascual-Leone & Johnson, 2021) to different brain regions and to well-attested brain networks that appear universal, such as the default mode network. Arsiladou distinguishes left from right hemisphere functions and relates key aspects of Pascual-Leone’s model such as the distinction between facilitating situations and misleading situations to aspects of visual brain function such as pro-saccades and anti-saccades (a pro-saccade is the automatic orienting reflex of moving the fovea to a novel visual cue that appears in the visual scene, while an anti-saccade is the automatic reflex of moving the fovea to a location opposite to the stimulus). Arsiladou’s analysis relies mainly on functional magnetic resonance imaging (fMRI) studies, which by their very nature do not tell us about causation. In fMRI, the blood oxygenation level or BOLD response, which takes time to peak, provides a biological correlate of neural activity at any given moment. Such correlations cannot tell us which aspects of neo-Piagetian theories are correct. Accordingly, the links to cognitive developmental neuroscience made in Arsalidou’s (2024) paper must be considered at the metaphorical level. For example, demonstrating right parietal activation in a simple number judgement task and more left-lateralized activation in a more complex calculation task is not a test of the right-left-right hemispheric dominance model that is outlined in Table 2. There is potentially some descriptive similarity, but whether there is a mechanistic or causal connection between these spatial changes in peak neural activation and the child’s mathematical behaviour remains to be discovered.Another frequent misconception in the cognitive neuroscience literature is that brain imaging studies can show that a particular developmental ability depends on the development or function of a particular neural structure. It is not clear that any currently available neuroimaging method could reveal causation of this type. Firstly, although BOLD responses in fMRI studies are typically analysed by setting a threshold for what counts as significant activation and even by using pre-determined regions of interest such as the frontal cortex, in reality there is continuous activation across the whole brain during any cognitive task, and so the isolation of particular structures is fraught with potential error. Further, neural structures adapt and change in response to environmental input, and logically any changes documented in a neural structure, such as the network of areas that are active during attentional tasks, could reflect cognitive-behavioural developments in attention rather than cause these changes. In her analysis of the development of attention, Rueda (2024) draws on theories of attention developed for adults and delineates the brain regions associated with different hypothesized adult attentional functions in order to create a framework for explaining the development of attention. Rueda explains clearly Posner’s influential adult work, which suggested three core functions of attention (alerting, selection, and executive control), and illustrates the neural networks associated with behavioural tasks that are intended to isolate these three functions. Rueda also explains how adult studies have suggested a popular mechanism called “biased competition” to explain how attention is allocated in real time, for example, when selecting what in the visual field to attend to. Rueda then provides an analysis at the descriptive level of how these insights from studies with adults could suggest a theoretical framework for the early development of attention, encapsulated in Figure 3. Rueda discusses some interesting data, providing a summary of clever neural paradigms that she and others have developed to study infant attention. However, it is not clear even at a descriptive level that mechanisms such as “biased competition” reflect the infant brain at work. At the mechanistic level, it is probably too early to assume that “age-related gains in behavioural and brain function efficiency are most likely related to changes in structural and functional connectivity observed with neuroimaging techniques.” The problem is that the causal direction may equally be from cognitive advances in the infant to changes in our neural measures.The paper by Crone and van Drunen (2024) acknowledges from the outset that correlations between activity in different brain regions and different kinds of cognitive tasks do not help us in explaining the causes of the development. Their paper focuses on the development of a self-concept. As they note, literally hundreds of fMRI studies have shown that a network of brain regions is consistently active when individuals have to mentalize (e.g., think about their intentions or the intentions of others), and this network has been labelled the “social brain.” Crone and van Drunen ask whether knowledge of the structures that are activated in this social brain network can help us to improve our theories of the development of a self-concept. Crone and van Drunen’s approach is to investigate whether brain activation can be meaningfully associated with prediction, in particular with predicting future individual differences in self-concept outcomes. This is a great idea, but the data that they offer only partly enable them to approach their goal. For example, they report that van Drunen et al. (2024) showed that shared environment had an influence on the development of structures such as the temporo-parietal junction (TPJ, a key element in the social brain network) between the ages of 7–14 years. Crone and van Drunen suggest that accordingly the TPJ might be highly sensitive to social experiences from middle childhood, thereby affecting the self-concept. This is one possibility, but it is not the only possibility. It might equally be that sensitivity to social experiences increases during middle childhood, and this affects the development of the TPJ. Despite the sophisticated brain imaging data discussed in their paper, it seems to me that Crone and van Drunen are not yet able to argue that structural and functional brain measures explain developmental trajectories of self-concept development. At the risk of sounding repetitive, parallel patterns between neural correlates and behavioural data attest to correlation, not causation. For example, Crone and van Drunen note that pro-social self-evaluation processes “rely on” the medial pre-frontal cortex and that an intervention that boosts self-esteem in adolescents also boosts the response in this neural structure. fMRI data do not provide evidence that a cognitive process “relies on” a neural structure; hence, converging neuroimaging methods are required to support this conclusion.The paper by Marshall (2024) relies more on developmental neuroscience, the study of the development of the nervous system, than cognitive developmental neuroscience. Marshall points out that brain function must always be considered in the context of the goal-directed activity of that brain’s owner. Too many enthusiasts for cognitive neuroimaging are stuck in the view of the brain as a computer that receives “inputs” from the external world, processes them (we believe we can measure these processes using our fMRI, fNIRS or EEG/MEG systems), and then delivers “outputs.” The recognition that the origin of all information in the brain is the activity of the infant or child herself is extremely important for theory and was recognized by both Piaget and Vygotsky, probably our most important developmental theorists. This perspective is also increasingly recognized by philosophers, particularly with relevance to understanding the development of the self (“Ambulo ergo sum,” rather than “Cognito ergo sum” [“I walk, therefore I am,” not “I think, therefore I am”]; O’Brien, 2007; 2015). Marshall does an accessible job of explaining how conceptual confusions about what genes do and about how we should conceptualize the nature of information have prevented knowledge from developmental neuroscience really impacting on theories of cognitive development. Some of Marshall’s most important points are also receiving recognition in adult cognitive neuroscience, such as the idea that cognitive behaviours are the output generated by the brain from its processing of information from the environment (also argued by Buzsaki, 2019: the “outside-in” view of brain function). Buzsaki proposed that the fundamental function of the brain is to produce actions as actions are the only way that an organism can make sense of sensory signals. Marshall’s suggestion is that a stronger focus on embodied cognition is the way forward if we are to enhance our developmental theories.However, even embodied cognition may not give us all the answers. Although quite sophisticated “embodied” models of cognition can now simulate the organism’s body and the physical world that it inhabits via sensors and robotic motor effectors, these models are still relatively simple (e.g., a simple “eye” that learns to track moving objects in the visual field, or a robotic hand that can pick up an egg without breaking the shell). Nevertheless, the fact that these “agent-based” models possess “bodies” that also provide input for neural learning is very important for developmental theory, and could be exciting in the future, particularly given the key role of action in Piaget’s theory of cognitive development. However, to truly inform developmental theory, the study of mechanism is key. A better understanding of the neural mechanisms underpinning action-based learning may bring us closer to understanding why development takes the path that we observe, particularly if detailed information about how actions increase knowledge can be studied in detail. Even so, the mental goal that is driving action in the first place is also a bodily action, albeit neurophysiological action such as neurotransmitter release. If we think of Vygotsky’s (1978) theory of cognitive development, then mechanisms like inner speech, pretending, and the imagination, generated by the child and then responded to psychologically by the child, play key developmental roles. These drivers of action are cognitive and internal; they are not sensory signals, but they are still actions of the body in the sense that even neurotransmitter release is an action of part of our body, our brain. How mechanisms like inner speech can be investigated using embodied cognition remains to be properly unpacked. In this sense, the field of developmental neuroscience still has a long way to go if it is to bring developmental theory to the forefront.In closing, I agree with many of the points made in these four interesting papers. The field needs to move beyond simply studying patterns of brain activation to studying brain mechanisms of information encoding and information processing, and using longitudinal and intervention designs is key. Longitudinal empirical studies at multiple levels of description are required, as attempted by both Rueda and Crone and van Drunen, and studies need to combine the assessment of individual differences in neural learning, sensory processing, cognitive processing and children’s behaviour. On current evidence, which is limited, it appears that children’s brains have essentially the same cortical and subcortical structures as the adult brain, carrying out essentially the same functions via the same mechanisms. The current contributions demonstrate a range of interesting approaches to meeting the Editors’ challenge of showing how the integration of neuroscience methods with developmental theory can advance the field’s understanding of developmental processes. Further, some areas of neuroscience are beginning to offer relevant tools for studying neural mechanisms that may cause individual differences in information encoding and thus help to explain why development takes the path that we observe. One example is oscillatory studies of the neuroscience of language processing (Goswami, 2020). As the study of brain mechanisms of information encoding and information processing continues to mature, it should be possible to progress even further our understanding of the development of cognitive representations such as the self-concept, cognitive processes such as attention, and developmental processes such as conceptual change.I would like to thank Drs. Susan Michelle Rivera and Caitlin Mahy for inviting me to contribute this commentary.The author declares no conflict of interest.The author has no funding to declare.Conceptualization, writing, and original draft.