Abstract
Behavior involves the ongoing interaction between an organism and its environment. One of the prevailing theories of adaptive behavior is that organisms are constantly making predictions about their future environmental stimuli. However, how they acquire that predictive information is still poorly understood. Two complementary mechanisms have been proposed: predictions are generated from an agent’s internal model of the world or predictions are extracted directly from the environmental stimulus. In this work, we demonstrate that predictive information, measured using bivariate mutual information, cannot distinguish between these two kinds of systems. Furthermore, we show that predictive information cannot distinguish between organisms that are adapted to their environments and random dynamical systems exposed to the same environment. To understand the role of predictive information in adaptive behavior, we need to be able to identify where it is generated. To do this, we decompose information transfer across the different components of the organism-environment system and track the flow of information in the system over time. To validate the proposed framework, we examined it on a set of computational models of idealized agent-environment systems. Analysis of the systems revealed three key insights. First, predictive information, when sourced from the environment, can be reflected in any agent irrespective of its ability to perform a task. Second, predictive information, when sourced from the nervous system, requires special dynamics acquired during the process of adapting to the environment. Third, the magnitude of predictive information in a system can be different for the same task if the environmental structure changes.
Highlights
Behavior involves the ongoing interaction between an organism and its environment
In the first part of this paper, we demonstrate that predictive information will generate indistinguishable results for systems that are at the two extremes of potential agent-environment interaction: a system whose only source of predictive information is the nervous system and a system whose only source of predictive information is the environmental stimuli
The study of predictive coding and its relevance to behavior has been studied from multiple perspectives in the literature: with regards to the source of information, predictive information can be generated by the neural network[5,6] and predictive information can be provided by the environment[7,22]
Summary
Of these systems is that in the CPGs the neural network drives its own activity and in the PP, the environment drives the neural network. This task allows us to specify the inherent structure in the environment by changing the distribution of objects whose attributes are compared making it especially suited for studying the influence of environmental structure on predictive information It involves providing the neural network with stimuli across three stages: cue, delay, and probe. These neural networks were not able to perform the relational categorization task (Fig. S2B), they encoded similar amounts of total predictive information as the trained neural networks (Fig. 4A) They encode the same amount of information about the probe during the cue stage (Fig. 4B). Unlike CPG and PP that were distinguished based on having different information sources, random and optimized neural networks in the relational categorization task have the same information sources Even under this condition, decomposing the total information across sources and unrolling over time helps distinguish them by revealing differences in the magnitude of information transferred from each source over time. Differences in environmental structure can result in significantly different amounts of predictive information encoded in neural networks without any behavioral differences
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
