Learning to Perceive the World as Probabilistic or Deterministic via Interaction With Others: A Neuro-Robotics Experiment

Abstract

We suggest that different behavior generation schemes, such as sensory reflex behavior and intentional proactive behavior, can be developed by a newly proposed dynamic neural network model, named stochastic multiple timescale recurrent neural network (S-MTRNN). The model learns to predict subsequent sensory inputs, generating both their means and their uncertainty levels in terms of variance (or inverse precision) by utilizing its multiple timescale property. This model was employed in robotics learning experiments in which one robot controlled by the S-MTRNN was required to interact with another robot under the condition of uncertainty about the other's behavior. The experimental results show that self-organized and sensory reflex behavior-based on probabilistic prediction-emerges when learning proceeds without a precise specification of initial conditions. In contrast, intentional proactive behavior with deterministic predictions emerges when precise initial conditions are available. The results also showed that, in situations where unanticipated behavior of the other robot was perceived, the behavioral context was revised adequately by adaptation of the internal neural dynamics to respond to sensory inputs during sensory reflex behavior generation. On the other hand, during intentional proactive behavior generation, an error regression scheme by which the internal neural activity was modified in the direction of minimizing prediction errors was needed for adequately revising the behavioral context. These results indicate that two different ways of treating uncertainty about perceptual events in learning, namely, probabilistic modeling and deterministic modeling, contribute to the development of different dynamic neuronal structures governing the two types of behavior generation schemes.

Bridging the Gap Between Probabilistic and Deterministic Models: A Simulation Study on a Variational Bayes Predictive Coding Recurrent Neural Network Model

“…To this end, we conducted a simulation experiment in which the VBP-RNN learned to predict/generate fluctuated temporal patterns containing probabilistic transitions between prototypical patterns. Consistent with Murata et al [13], results showed that the different weighting arbitrates between two extremes at which the model develops either a deterministic dynamic structure or a probabilistic one. Analysis on simulation results clarifies how the degree of generalization in learning as well as the strength of the top-down intentionality in generating patterns changes from one extreme to another.…”

“…In this direction, Murata et al [13] developed a predictive coding-type stochastic RNN model inspired by the free energy minimization principle [7]. This model learns to predict the mean and variance of sensory input for each next time step of multiple perceptual sequences, mapping from its current latent state.…”

“…In the case of strong initial state dependency, deterministic chaos is dominantly developed, while stochastic dynamics develops by estimating larger variance in the case of weak initial state dependency. However, the model in [13] suffers a considerable drawback. It cannot estimate variance for latent variables (the context units).…”

A Novel Predictive-Coding-Inspired Variational RNN Model for Online Prediction and Recognition

“…The error regression process was implemented in deterministic RNNs, and it was shown how it could help the generalization capability of those models (Tani & Ito, 2003;Murata et al, 2017;Ahmadi & Tani, 2017b). This testing process through error regression bears similarities to, and is inspired by, predictive coding.…”

Learning, planning, and control in a monolithic neural event inference architecture

“…Related work with RNNs has implemented hierarchical RNN architectures that develop symbol-like encodings in the deeper RNN layer via gradient descent (Tani, 2003). Later, the process was termed an error regression scheme and was closely related to the free energy principle (Murata et al, 2017). Related work has also been put forward when setting internal hidden states -often referred to as parametric bias neurons -to induce particular behavioral primitives and sequences thereof by suitably trained hierarchical RNN architectures (Arie et al, 2009;Tani, 1996;.…”