Future directions meeting – abstracts

Day 1 Monday March 26th

9 – 9.30 Introduction and presentation of attendants

9 – 10.30, Session 1: Introduction to the Workshop — discussants: Tobias Heed (Bielefeld) & Kevin O’Regan (Paris)

Starting as early as the behaviourists and going through Piaget till the present day, there is wide consensus about  the idea that an infant learns about the world and itself by making movements and observing the resultant effects. Ideomotor theory, for instance, assumes that after early development, actions are planned by recalling the (learned) movements that will lead to a desired outcome. So the idea of sensorimotor contingency (SMC) is not new. But despite many years of research, our understanding of how SMC’s unfold into an “understanding” of our sensory experience or into a “representation” of the body is a sketch at best.

At this meeting, we invite attendees to envision, and possibly chart concrete goals about how research could tackle open questions and develop ideas that allow testing the SMC idea in ways that allow going beyond the current sketch, towards explicitly formulated theory. The different sessions we have proposed reflect areas we think all contribute to the SMC idea, or are somehow based on it, and so each of these areas will hopefully shed a different light on our debate.

–       Session 2 looks at the relationship of SMC and space, and ask whether “maps” and/or “statistics” are useful concepts to grasp SMC

–       Session 3 discusses whether research in adult populations – which presumably is easier than research in children – allows drawing valid inferences about early development

–       Session 4 takes a look at whether recent methodological advances might be key in understanding SMC

–       Session 5 addresses the real life difficulties of infant and child research, how they limit understanding SMC, and how they might be overcome

–       Session 6 asks whether there is merit in aligning sensory and motor development, based on the implication of SMC that the two areas are inseparable

–       Session 7 moves further backward in time, asking how genetic programs may be the basis for initial unfolding of SMC, and what starting point gene-behavior interactions in the womb create for SMC after birth

–       We will attempt to pool the outcome of our discussions in Session 8.

Here is an example of how current theory and experimentation often does not sufficiently elaborate the SMC approach. A sensorimotor view implies that an embryo and/or infant cannot differentiate external (e.g. tactile) from internal (e.g. some organ) stimulation: it cannot at first “feel” touch “on” its body or “at” a particular location. Genetically implemented response tendencies, such as reflexes, are part of automatic sensorimotor loops that cannot yet have been structured with respect to each other. Similar arguments can be made about vision, hearing, etc. If this is true, the question arises of how “sense”, or structure, emerges from ongoing experience, and of how they lead to purposeful action. What is the relationship between continuous SMC learning and sudden developmental steps and “aha” moments? Why does sensitivity to contingencies change the way it does over development? Which among myriads of sensorimotor contingencies does the baby attend to at any moment and why? How, and why, does experience lead to the organism separating body, self, and world? How do classes of related contingencies come together to form what adults call sensory modalities? How are sensory modalities coordinated together through what appears to be spatial coordinate frames? If conceptual knowledge is based on SMC’s, then how can we develop abstract concepts?

Our impression is that many of us have some intuitions about how these questions could be answered. Yet, research has often asked narrow and superficial questions, for example showing that infants are sensitive to sensorimotor contingencies, characterising some temporal, featural, and spatial characteristics of contingencies that the baby is sensitive to at different ages. However, we propose that real understanding requires a more dedicated theoretical approach. In this discussion session, we will solicit common ideas shared by attendees, and sketch open questions.

11 – 12.30, Session 2: Brain mechanisms: maps vs. statistical inference (invited discussant: Suliann Ben Hamed, CNRS Lyon, France)

The notion of sensory and motor maps has been extremely instrumental in describing how the brain organizes functions. From an anatomical point of view these maps conveniently match well-defined anatomical territories characterized by specific layer organizations and specific anatomical (hard wired) connectivity with other cortical (or sub-cortical) territories. From a functional point of view, specific neuronal computations have been described in each of these territories, leading to the general idea that functions arise from specific convergence of neuronal inputs and specific local neuronal computations. Overall, this has led to comprehensive generative models of major brain functions. However, this approach had also led to an ever growing level of unsatisfactory complexity that now calls for a reassessment of how these maps should be viewed.

Focusing on sensory and sensori-motor maps and coordinate transformation, I intend to trigger a discussion on the definition and validity of the notion of functional maps: what is their actual granularity? what do areal/map boundaries actually mean? Are functions implemented in areas or in networks? how? I will contrast this with the idea that sensorimotor processes can be viewed as statistical (spatial) inference processes the specificities of which (in how they relate to functional maps) are determined by the weighing of both incoming sensory (bottom-up) information and internal (top-down) information. I will extend this discussion to a statistical and a dynamic neuronal view of body knowledge and coordinate transformation.

In a last step, I will discuss the fact that body anchored representations are dynamic and flexible, that is are changing online as opposed to changing under developmental and learning processes. I will draw on examples from the study of peripersonal space and attention and I will raise the question of whether sensorimotor contingencies can be viewed independently of other cognitive functions such as emotions, social cognition and motivation.
All throughout the discussion, methodological considerations and innovations and the identification of knowledge gaps will be addressed. I except this discussion, thought centered onto adult brain mechanisms to raises a rich discussion in relation with the developmental perspective.

14.30 – 16.00, Session 3: Generality of the sensorimotor contingencies approach as a central learning principle across the life span (invited discussant: Olivier Collignon, University of Louvain, Belgium)

In this session I intend to trigger a discussion about how experience early in life has a particular value in shaping how the mind and brain develop to process information. During early infancy, sensing the world around us not only provides the infants with sensory information of the environment, but it also sculpts the very structure and function of the brain. The particular value of experience early in life notably relates to the plastic state of the brain at this developmental period. This however does not mean that the brain is a blank slate waiting to be imprinted by the organization of the outside world. We are born with native predisposition to process information and this intrinsic structure will interact and constrain the way the mind and brain will evolve in relation to the particular sensorimotor contingencies each individual experiences. This framework should help us to pave the way for a stimulating discussion about whether the child is father of the man.

16.30 – 18.00, Session 4: From observing behavior to understanding body knowledge development (invited discussant: Aldo Faisal, Imperial College London, UK)


Day 2 Tuesday March 27th

9 – 10.30, Session 5: Sensorimotor contingencies in infants: “real life” research (invited discussant: Andrew Bremner, Goldsmiths University of London, UK)

There is a long history of interpretational difficulties in research with human infants. An explosion of research using looking duration methods for understanding infants’ abilities from the 1960s onwards led to many reports of early, and often surprising, competence even in infants of a few days of age. I will put forward a view that there are four principal difficulties which have constrained the progress which we can make with regard to understanding the development of perception and representation through infant behavioural methods. The first is that infants are intolerant of long testing sessions and boring stimuli (thus compromising control and design). The second is that null findings with infant behavioural measures are rarely easy to interpret (does this indicate an absence of the perceptual/cognitive ability in question, an absence of the behaviour which would index it, or were the babies paying no heed to assumed demands characteristics [sic]?). The third is that researchers rarely agree about what cognitive abilities behavioural measures reveal. The fourth is that babies do not seem to develop behaviourally in any kind of universal manner; individual differences in the development of sensorimotor abilities are substantial.

In this discussion session, I will start by arguing that behavioural methods with infants are crucial for understanding the sensorimotor millieu in which infants learn about their worlds. After all, without an understanding of the behaviours which infants engage in with their worlds, how would we know what options are available for them to learn sensorimotor contingencies? However, in order to address the problems mentioned above, I advocate, as have a number of others, the use of convergent methods (combining a range of physiological and behavioural methods addressing the same putative ability) as a means of refining our understanding of the nature of infant competence at any given age, and in any given individual. As a way of characterising how development proceed, I also increasingly think that it is important to measure changes against other proxies for development than age (in months).

Lastly, and perhaps of key importance to the sensorimotor account, we need to develop better, and more coherent approaches to understanding the causal processes of development. Training methods with infants as well as convergent input from research on developmental disorders and sensory deprivation is crucial in this regard.

11 – 12.30, Session 6: Aligning the “sensory” and the “motor” in “sensorimotor” (invited discussant: Carina de Klerk, Birkbeck College London, UK)

From the moment they come into the world, infants are surrounded by people who are performing actions that they cannot yet perform themselves. Not only do infants learn to master a lot of these actions in the first few months and years of life, they also get better at processing and predicting these actions when they are performed by others. In recent years, there has been a renewed interest in the idea that these abilities do not merely happen to develop in parallel, but rather that motor and sensory development are closely intertwined. In my talk I will first present evidence for this idea, and discuss theories that have been proposed to explain the close link between motor and sensory processes (e.g. the sensorimotor learning accounts (Heyes, 2010; Keysers & Perrett, 2004)). While these theories mainly aim to explain the link between the execution of one’s own actions and the observation of others’ actions, similar mechanisms may also play an important role in linking up sensory and motor representations relating to one’s own actions, hereby supporting the development of body knowledge. If this is true, a fruitful avenue for future research would be to investigate the alignment of sensory and motor milestones and their impact on the infant’s developing body representations. I will discuss whether we can define such milestones, what methods we might use to observe them, and the difficulties that one might encounter when using such an approach. Finally, I will discuss the benefits of studying both typical and (a) typical sensorimotor development trajectories, as well as the benefits of training studies in which the different aspects of motor and sensory experience are manipulated to disentangle the effects of sensory, motor, and sensorimotor experience on the development of body knowledge.

14.30- 16.00, Session 7: gene-behavior interaction before birth (invited discussant: Nadja Reissland, Durham University, UK

My presentation will centre on the following question: Can we gain a better understanding of fetal behaviour and its relevance in creating initial body knowledge?  Current epigenetic research indicates that the genome is not fixed, with the editor of the Journal of Clinical Epigenetics writing: “Genome-wide association studies are no longer expected to yield the markers to fully explain the heritable susceptibility observed for many complex diseases”. (Rots 2015).  And Black et al (2016) asserted: “for many common conditions with familial components, such as schizophrenia, the majority of their heritability remains “hidden,” most likely in rare variants invisible to Genome Wide Association Studies”.

The current thinking around epigenetics suggests that there is no clear link between the (invariant) genome and behavioural outcome; and therefore I am  going to focus my discussion on how the fetus creates an initial body map.

To date, the development of the fetus has been measured in terms of physical development but little has been done to understand other changes – such as the development of awareness. This is a topic that is currently being debated in both medical and political spheres, and forms part of a governmental enquiry into pain experience of the fetus. When does the fetus experiences pain? And, if it does, can the pain experience mitigated by maternal state, such as depression, stress or anxiety?

In order to experience pain, the fetus must have some level of awareness. How can we test this awareness? I suggest that there are several ways to examine this question, by longitudinal observation and experimental interventions with various stimuli including sound, light, taste, and olfaction.

A second way in which we can answer the question of how the fetus creates initial body knowledge is by examining touch behaviours in singleton versus twins.

Touch is par excellence a method to engender initial body knowledge. This body knowledge changes depending on whether you are a singleton or twin.

The singleton fetus will only be able to touch him or herself and the womb environment; the twin will already experience the self-touch versus other touch difference prenatally. I shall discuss this topic illustrating it with ultrasound images.

Limitations to fetal observational and experimental non-invasive studies concern mainly the imaging equipment currently available, the cost of imaging and limitations imposed (by rules of safeguarding) on time allowed for nonmedical imaging (British Medical Ultrasound Society) as well as the mother’s physical and psychological state.


16.30 -18.00, Session 8: Wrapping up: pulling the strings together and defining new directions (Tobias Heed, Bielefeld University, Germany)

Overview of the five sessions’ main results, discussion of conclusions and possible next steps to take in the field