Sensory attenuation develops as a result of sensorimotor experience

The brain attenuates its responses to self-produced exteroceptions (e.g., we cannot tickle ourselves). Is this phenomenon, called sensory attenuation, enabled innately, or is it acquired through learning? To explore the latter possibility, we created a neural network model consisting of sensory (proprioceptive and exteroceptive), association, and executive areas. A simulated robot controlled by the network learned to acquire motor patterns with self-produced or externally produced exteroceptive feedback. We found that the robot first increased responses in sensory and association areas for both self-produced and externally produced conditions in the early stage of learning, but then, gradually it attenuated responses in sensory areas only for self-produced conditions. The robot spontaneously acquired a capacity to switch (attenuate or amplify) responses in sensory areas depending on the conditions by switching the neural state of the executive area. This suggests that proactive control of sensory-information flow inside the network was self-organized through learning. We also found that innate alterations in the modulation of sensory-information flow induced some characteristics analogous to schizophrenia and autism spectrum disorder. This study provides a novel perspective on neural mechanisms underlying perceptual phenomena and psychiatric disorders. Introduction The brain couples its structure with the outside world via sensorimotor experiences (1). A posteriori development of neural processing gradually forms our perceptual phenomena, with a priori nature determined by genes. It yields well-defined self-experience and helps to confidently situate self in relation to others. In general, we face two primary types of sensorimotor experience, self-movements and sensory events in the outside 1 ar X iv :2 11 1. 02 66 6v 1 [ qbi o. N C ] 4 N ov 2 02 1 Brain model develops sensory attenuation world, which may or may not be correlated. Recognition of the difference is thought to underlie self-other distinction or sense of self (2, 3), but the difference cannot be known a priori. The brain may acquire the capacity to modulate neural responses via sensorimotor learning, depending on the condition. A phenomenon called sensory attenuation, is recognized as one of the bases of the sense of self, especially the sense of agency (2, 4). Sensory attenuation refers to an experience in which an exteroception produced by a self-movement is less salient than one produced externally. A perfect example is the difficulty of tickling ourselves. In addition, the ability to ignore visual changes during eye or head movements is thought to help maintain stability of the visual scene. At the neural level, sensory attenuation is observed as reduced brain responses, such as blood oxygen level-dependent (BOLD) responses, especially in sensory areas (5–7). A line of research shows that sensory attenuation is associated with predicted sensorimotor correlation and is diminished by temporal or spacial mismatch between a movement and the resultant exteroception (e.g., a temporal delay) (8–10). There are some suggestions of underlying neural functions, such as an efference copy of the motor command (8,11) and neuromodulation (e.g., dopaminergic transmission) (12). However, despite intensive work for decades, a fundamental question remains unexamined: Is sensory attenuation enabled innately or is it acquired through learning? The latter possibility remains almost entirely unexplored, although several studies have shown increased sensory attenuation with age in adults (13, 14). Here, we provide a novel mechanistic explanation, suggesting that sensory attenuation can be self-organized through learning. We focus on the following factors. If a self-movement and a sensory event in the outside world are not correlated, the resultant proprioception and exteroception occur separately. The brain may efficiently use individual sensory areas to process these individual sensations. On the other hand, if the proprioception and exteroception are correlated, they can be perceived as coupled, rather than individual sensations. In that case, it is reasonable to process the sensorimotor coupling using an association area, and neural responses in individual sensory areas can be attenuated. That is, through sensorimotor learning, the brain may develop the capacity to modulate sensory-information flow and processing inside the hierarchical neural networks, depending on the conditions. To test this hypothesis, we conducted a robot simulation experiment using a variational recurrent neural network model based on Bayesian brain theory or free-energy minimization (1,15–18). In addition, we show that some sets of characteristics, analogous to schizophrenia and autism spectrum disorder (ASD), can result from alterations in this modulatory function, induced by manipulation of an innate network parameter. Results Computational model Our daily sensorimotor behavior generally has some sort of regularity (e.g., a set-point of body posture and dynamic movement patterns) and randomness (e.g., free movement with fluctuations) (19). Therefore, we considered how a robot agent controlled