We present a computational model of the mechanisms that may determine infant behavior in the "mobile paradigm." This paradigm has been used in developmental psychology to explore how infants learn the sensory effects of their actions. In this paradigm, a mobile (an articulated and movable object hanging above an infant's crib) is connected to one of the infant's limbs, prompting the infant to preferentially move that "connected" limb. This ability to detect a "sensorimotor contingency" is considered to be a foundational cognitive ability in development. To understand how infants learn sensorimotor contingencies, we built a model that attempts to replicate infant behavior. Our model incorporates a neural network, action-outcome prediction, exploration, motor noise, preferred activity level, and biologically inspired motor control. We find that simulations with our model replicate the classic findings in the literature showing preferential movement of the connected limb. An interesting observation is that the model sometimes exhibits a burst of movement after the mobile is disconnected, shedding light on a similar occasional finding in infants. In addition to these general findings, the simulations also replicate data from two recent more detailed studies using a connection with the mobile that was either gradual or all-or-none. A series of ablation studies further shows that the inclusion of mechanisms of action-outcome prediction, exploration, motor noise, and biologically inspired motor control was essential for the model to correctly replicate infant behavior. This suggests that these components are also involved in infant sensorimotor learning.

Learning sensorimotor contingencies—that is, the link between one’s actions and their sensory effects—is fundamental to developing body knowledge, understanding causality, and developing a sense of agency. In developmental psychology, this process is classically studied using the mobile paradigm, where infants learn that movement of a limb causes motion of a connected mobile. To expand our understanding of how infants learn this, we tested an embodied computational model that learns through two biologically inspired mechanisms: prediction and curiosity. Implemented on the child-sized iCub humanoid robot interacting with a mobile, the model detected sensorimotor contingencies across several experimental conditions using a variety of movement strategies. Our findings suggest that contingency learning cannot be captured by a single behavioral metric, such as the amount of movement, but instead emerges through a spectrum of exploratory behaviors. Analysis of the robot’s internal activity reveals that these behaviors emerge from the dynamic trade-off between prediction and curiosity—between exploitation and exploration. Our work provides a biologically motivated, physically embodied model of sensorimotor interaction that connects theories of infant learning with robotic implementations. The results allow us to generate testable hypotheses for developmental research and to inform the design of autonomous learning systems.

This longitudinal study investigated the effect of experience with tactile stimulation on infants' ability to reach to targets on the body, an important adaptive skill. Infants were provided weekly tactile stimulation on eight body locations from 4 to 8 months of age (N = 11), comparing their ability to reach to the body to infants in a control group who did not receive stimulation (N = 10). Infants who received stimulation were more likely to successfully reach targets on the body than controls by 7 months of age. These findings indicate that tactile stimulation facilitates the development of reaching to the body by allowing infants to explore the sensorimotor correlations emerging from the stimulation.

Specifying surface reflectances in a simple and perceptually informative way would be beneficial for many areas of research and application. We assessed whether a 3×3 matrix may be used to approximate how a surface reflectance modulates the sensory color signal across illuminants. We tested whether observers could discriminate between the model’s approximate and accurate spectral renderings of hyperspectral images under narrowband and naturalistic, broadband illuminants for eight hue directions. Discriminating the approximate from the spectral rendering was possible with narrowband, but almost never with broadband illuminants. These results suggest that our model specifies the sensory information of reflectances across naturalistic illuminants with high fidelity, and with lower computational cost than spectral rendering.

Two very fundamental aspects of phenomenal experiences underline the fact that they seem to have "something it's like." One aspect is the fact that experiences have a locus: they Can seem "external" (perceptual), "internal" (interoceptive, bodily or emotional) or "mental." A second fundamental aspect is the imposingness of experiences. They can seem "present" to us in different ways, sometimes seeming displayed before us with "spatio-temporal presence." Both these aspects of "what it's like" can be identified with the degree to which we can voluntarily control what we are doing when we engage in an experience. The external/internal/mental dimension is determined by how our voluntary bodily actions can influence the sensorimotor flow of information. The degree of imposingness of experiences and their "spatio-temporal presence" Is determined by how our voluntary actions are impeded or assisted by innate, attention-grabbing mechanisms. By elucidating these two most fundamental aspects of "what it's like," and taken together with prior work on inter- and intra-modal differences in experiences, this article suggests a path toward a scientific theory of the "hard problem" of phenomenal consciousness, explaining why experiences feel like something rather than feeling like nothing.

This study investigated the potential for the development of novel perceptual experiences through sustained training with a sensory augmentation device. We developed (1) a new geomagnetic sensory augmentation device, the NaviEar, and (2) a battery of tests for automaticity in the use of the device. The NaviEar translates head direction toward north into continuous sound according to a “wind coding” principle. To facilitate automatization of use, its design is informed by considerations of the embodiment of spatial orientation and multi-sensory integration, and it uses a sensory coding scheme derived from means for auditory perception of wind direction that is common in sailing because it is easy to understand and use. The test battery assesses different effects of automaticity (interference, rigidity of responses, and dynamic integration) assuming that automaticity is a necessary criterion to show the emergence of perceptual feel, that is, an augmented experience with perceptual phenomenal quality. We measured performance in simple training tasks, administered the tests for automaticity, and assessed subjective reports through a questionnaire. Results suggest that the NaviEar is easy and comfortable to use and has a potential for applications in real-world situations. Despite high usability, however, a 5-day training with the NaviEar did not reach levels of automaticity that are indicative of perceptual feel. We propose that the test battery for automaticity may be used as a benchmark test for iterative research on perceptual experiences in sensory augmentation and sensory substitution.

Specifying surface reflectances in a simple and perceptually informative way would be beneficial for many areas of research and application. We assessed whether a 3 x 3 matrix may be used to approximate how a surface reflectance modulates the sensory colour signal across illuminants. We tested whether observers could discriminate between the model’s approximate and accurate spectral renderings of hyperspectral images under narrowband and naturalistic, broadband illuminants for 8 hue directions. Discriminating the approximate from the spectral rendering was possible with narrowband, but almost never with broadband illuminants. These results suggest that our model specifies the sensory information of reflectances across naturalistic illuminants with high fidelity, and with lower computational cost than spectral rendering.

This article accounts for two of the most fundamental aspects of “what it’s like” to have an experience, thereby showing a path towards solving the “meta”-”hard problem” of phenomenal consciousness. One fundamental aspect of “what it’s like” to have an experience is that experiences have qualities rather than no qualities. Among the qualities, the most fundamental are the fact that experiences are classifiable as being of an “external” (perceptual), an “internal” (interoceptive, bodily or emotional) or of a “mental” kind. A second fundamental aspect of “what it’s like” seems to be that experiences impose themselves upon us: they seem “present” to us -- they feel like something rather than feeling like nothing. Both these aspects of “what it’s like” can be accounted for in terms of the degree to which we can voluntarily control what we are doing when we engage in an experience. The external/internal/mental dimension is determined by how our voluntary bodily actions can influence sensorimotor flow of information. The imposingness of experiences and their “spatio-temporal presence” is determined by how our voluntary actions are impeded or assisted by innate, attention-grabbing mechanisms.

This summary of the sensorimotor theory of phenomenal consciousness explains the purpose of the theory, emphasizes its metaphysical differences with most other current theories, gives links to overviews of supporting evidence, discusses frequent misunderstandings of the theory, and describes the future program of work.

Infants start to use a spoon for self-feeding at the end of the first year of life, but usually do not use unfamiliar tools to solve problems before the age of 2 years. We investigated to what extent 18-month-old infants who are familiar with using a spoon for self-feeding are able to generalize this tool-use ability to retrieve a distant object. We tested 46 infants with different retrieval tasks, varying the tool (rake or spoon) and the target (toy or food). The tasks were presented in a priori descending order of difficulty: rake–toy condition, then either spoon–toy or rake–food, and finally spoon–food. Then, the same conditions were presented in reverse order to assess the transfer abilities from the easiest condition to the most difficult retrieval task. Spontaneously, 18-month-old infants performed the retrieval tasks better with the familiar tool, the easiest task being when the spoon was associated with food. Moreover, the transfer results show that being able to use a familiar tool in an unusual context seems necessary and sufficient for subsequent transfer to an unfamiliar tool in the unusual context, and that early and repetitive training of self-feeding with a spoon plays a positive role in later tool use.

When leaving the aquatic constrained environment of the womb, newborns are thrown into the world with essentially new laws and regularities that govern their interactions with the environment. Here, we study how spontaneous self-contacts can provide material for learning implicit models of the body and its action possibilities in the environment. Specifically, we investigate the space of only somatosensory (tactile and proprioceptive) activations during self-touch configurations in a simple model agent. Using biologically motivated overlapping receptive fields in these modalities, a variational autoencoder (VAE) in a denoising framework is trained on these inputs. The denoising properties of the VAE can be exploited to fill in the missing information. In particular, if tactile stimulation is provided on a single body part, the model provides a configuration that is closer to a previously experienced self-contact configuration. Iterative passes through the VAE reconstructions create a control loop that brings about reaching for stimuli on the body. Furthermore, due to the generative properties of the model, previously unsampled proprioceptive-tactile configurations can also be achieved. In the future, we will seek a closer comparison with empirical data on the kinematics of spontaneous self-touch in infants and the results of reaching for stimuli on the body.

This literature review examines how babies’ body know-how develops during the first year of life. It surveys studies describing this development through the exploration of the body and of the physical environment. This early development may help babies acquire a sense of agency and a sense of body ownership. The development of body know-how, as a precursor to more in-depth knowledge of the body and of the self, may play an essential role in children's socio- cognitive and psychomotor development.

Infants’ ability to monitor “sensorimotor contingencies,” i.e., the sensory effects of their own actions, is an important mechanism underlying learning. One method that has been used to investigate this is the “mobile paradigm,” in which a mobile above an infant’s crib is activated by motion of one of the infant’s limbs. Although successfully used in numerous experiments performed in infants’ homes to investigate memory and other types of learning, the paradigm seems less robust for demonstrating sensitivity to sensorimotor contingencies when used in the laboratory. One purpose of the present work was to show that certain changes to the mobile paradigm would make it easier for infants to show their sensitivity to the contingency in the lab. In particular, we used proximal stimulation on infants’ wrists instead of the usual mobile, and our stimulation was coincident with the limbs that caused it. Our stimulation was either on or off, i.e., not modulated by the amount the infant moved. Finally, we used a “shaping” procedure to help the infant discover the contingency. In addition to these changes in the paradigm, by analyzing infants’ limb activity at 10-s resolution instead of the usual 1-min resolution, we were able to show that infants’ sensitivity to the contingency became apparent already within the first minute of establishment of the contingency. Finally, we showed how two alternate measures of sensitivity to contingency based on probability of repeated movements and on “stop and go” motion strategies may be of interest for future work.

In order to benefit from the exploration of their body and their physical and social environment, infants need to detect sensorimotor contingencies linking their actions to sensory feedback. This ability, which seems to be present in babies from birth and even in utero, has been widely used by researchers in their study of early development. However, a careful look at the literature, particularly recent literature, suggests that babies may not be uniformly sensitive to all sensorimotor contingencies. This literature review examines in detail the mechanism of sensorimotor contingency detection in infants before the age of one year. Four aspects of sensorimotor contingency detection are considered: characteristics of action and feedback, contingency parameters, exposure conditions, and inter-individual differences. For each topic we highlight what favours and what hinders the detection of sensorimotor contingencies in infants. Our review also demonstrates the limitations of our knowledge about sensorimotor contingency detection. We advocate the importance of making progress in this field at a time when sensorimotor contingency detection is of major interest in developmental robotics and artificial intelligence.

The ability to perceive and use the body parts in an organised and differentiated manner is a precursor of body knowledge in infancy. To acquire this ability, the infant’s brain might explore the perceptual consequences of its bodily actions. Undifferentiated body movements would gradually be replaced by more precise actions. Only a very few papers have tested this “global-to-local” hypothesis and none of them have so far been replicated. In this study, we assessed arm differentiation in 4-, 6- and 8-month-old infants using a new contingency detection task in which infants have to detect a contingency between one of their arms’ activity and an audiovisual stimulus on a screen. We found that 4- to 8-month-old infants seem able to differentiate their arms. However, surprisingly, we were not able to show a developmental trend in arm differentiation between 4 and 8 months of age.

Cette revue de la litterature propose d’examiner de quelle maniere le savoir-faire corporel du bebe s’affine au cours de la premiere annee de vie, en decrivant ce developpement a travers l’exploration du corps et l’exploration de l’environnement physique. Ce developpement precoce pourrait participer a l’acquisition par le bebe d’un sens de l’agentivite (sense of agency) et d’un sens du corps propre (body ownership). Le developpement du savoir-faire corporel, par son statut de precurseur d’une connaissance plus approfondie du corps et de soi, jouerait un role essentiel dans le developpement sociocognitif et psychomoteur de l’enfant.

Developmental psychology experiments on tool use show that infants’ capacity to use a rake-like tool to retrieve a toy arises quite suddenly around 18 months. We use a developmental-robotics model to propose and test two alternative hypotheses to explain this conundrum. Both hypotheses rely on the assumptions that tool use involves goal-directed behavior processes guided by the goal of retrieving the toy, and that “understanding how to use a tool” means acquiring the capacity to assemble a sequence of actions to accomplish the goal (e.g., to “hook” and then “retrieve” the toy). The first hypothesis is that the tool-use ability emerges when the infant develops enough planning capabilities. The second hypothesis is that the ability emerges when the infant’s intrinsic motivation system develops and makes playing with a couple of objects interesting enough so that the infant plays with objects similar to the tool at home and thus acquires the actions needed to retrieve the toy in the lab. These hypotheses are tested through a neural-network architecture controlling a simulated humanoid robot tested with the tool-rake task. Given the assumptions made in the model, the results show that both hypotheses can reproduce the average behavior of infants but only the intrinsic-motivation hypothesis can reproduce the sudden tool-use improvement.

Cette revue de la littérature propose d’examiner de quelle manière le savoir-faire corporel du bébé s’affine au cours de la première année de vie, en décrivant ce développement à travers l’exploration du corps et l’exploration de l’environnement physique. Ce développement précoce pourrait participer à l’acquisition par le bébé d’un sens de l’agentivité ( sense of agency ) et d’un sens du corps propre ( body ownership ). Le développement du savoir-faire corporel, par son statut de précurseur d’une connaissance plus approfondie du corps et de soi, jouerait un rôle essentiel dans le développement sociocognitif et psychomoteur de l’enfant.

Doerig et al. evaluate how current empirical theories approach access consciousness, but they neglect how they approach phenomenal consciousness - probably because most theories don't deal with phenomenal consciousness at all. One exception is the sensorimotor theory, but Doerig et al. did not evaluate it as being directed to phenomenal consciousness.

Adults are capable of very fine motor skills whereas newborn babies’ motions are less accurately adjusted to the environment. It has been suggested that babies are sensitive to sensorimotor contingencies so they can acquire their body knowhow by gradually linking each body movement to its perceptual consequences. The research we pursued in the team is part of this theoretical framework. We use behavioural measurements to study how babies refine their body knowhow over time.During my internship, we studied arm differentiation in infants of age 6 months. An artificial contingency was established between the movements of one of the babies’ arms and the appearance of visual and auditory stimuli on both of their arms. My goal was to develop analytical tools to assess if babies detect the contingency (i.e. if they realize that they caused the occurrence of the stimuli). I tried to reproduce the probabilistic methodology developed by J. Watson in his experiments with 4month old babies. I could not obtain reliable results and so pursued my investigations. I adapted Watson’s analytical tools to create a binary indicator measuring the success of babies at the individual level. I showed that babies can differentiate between a situation where without doubt they have no control and a situation where they could be the cause of the stimulus. However, because babies who tried to test the contingency behaved similarly in both the test and the control group I can not ascertain that babies from the contingent group understood that they triggered the contingency.