New Labels for Old Ideas: Predictive Processing and the Interpretation of Neural Signals

Abstract

Philosophical proponents of predictive processing cast the novelty of predictive models of perception in terms of differences in the functional role and information content of neural signals. However, they fail to provide constraints on how the crucial semantic mapping from signals to their informational contents is determined. Beyond a novel interpretative gloss on neural signals, they have little new to say about the causal structure of the system, or even what statistical information is carried by the signals.That means that the predictive framework for perception can be relabeled in traditional, non-predictive terms, with no empirical consequences relevant to existing or future data. To the extent that neuroscientific research based on predictive processing is both innovative and productive, it will be due to the framework's suggestive heuristic effects, or perhaps auxiliary empirical claims about implementation, rather than a difference in the information-processing structure that it describes.

Predictive processing as a systematic basis for identifying the neural correlates of consciousness

“…In this discussion we have focused on the promising explanatory and unificatory aspects of PP. However, PP, although increasingly pervasive, is certainly not universally accepted and there are conflicting findings and perspectives, driving ongoing debate (e.g., Alilović et al, 2019;Cao, 2020;Li & Ma, 2020;Litwin & Miłkowski, 2020;Rahnev & Denison, 2018). For example, one recurring challenge to PP is that worry that an imperative to minimize sensory prediction errors will lead to agents that do nothing at all -the so-called 'dark room problem' (Friston et al, 2012;Sun & Firestone, 2020a, 2020b.…”

Can expectation suppression be explained by reduced attention to predictable stimuli?

“…Over the last decades, theories of predictive coding became more and more popular and expectation suppression is an observation that has frequently been interpreted as supporting such theories. Although we agree that reduced prediction error coding provides an appealing explanation for reduced responses to expected stimuli, we here want to join others in requesting that predictive coding should not be accepted as an ‘a-priori truth’ and that we should remain open to considering alternative explanations to prevent this theoretical framework from turning into a ‘just-so story’ that is able to accommodate any empirical finding (Bowers & Davis, 2012; Cao, 2020; Heilbron & Chait, 2017; Walsh et al, 2020). Note, however, that the same issue applies to the concept of attention, which appears to be equally ill-defined (Hommel et al, 2019).…”

“…A current debate around theories of predictive coding is whether they provide new constraints on brain function that have not already been provided by traditional models (Cao, 2020). In this perspective we contribute to this debate by discussing the possibility that expectation suppression, an empirical finding frequently interpreted as support for predictive coding theories, can be accounted for by traditional models of attention.…”

“…'a-priori truth' and that we should remain open to considering alternative explanations to prevent this theoretical framework from turning into a 'just-so story' that is able to accommodate any empirical finding(Bowers & Davis, 2012;Cao, 2020; Heilbron & Predictive-coding based explanation (PBE, left) attention increases the responsiveness of neurons sensitive to the discrepancy of the sensory input and the expected stimulus. Attention-based explanation (ABE, right) explains expectation suppression as resulting from a reduced stimulus saliency of expected stimuli.…”

You can't always get what you want

“…Rosa Cao (2020) has recently cautioned that the underdetermination of the algorithmic solutions by the empirical data goes beyond different varieties of the same computational scheme and that the same neuroanatomical evidence might be equally consistent with PP as well as other, more traditional message-passing algorithms. She shows that -as far as the flow of information is concerned -PP models of neural signalling can be "relabelled in traditional, non-predictive terms, with no empirical consequences relevant to existing or future data" (Cao, 2020, p. 517).…”