Table 1.

Structural and functional aspects of the plant vascular system that may be explained under a free-energy (predictive processing) formulation.

domain	prediction
anatomy and connectivity: explains the hierarchical deployment of plant vascular bundles, architectures with forward and backward (bidirectional) connections [1,54,55,57,58]	—hierarchical vascular organization —distinct vascular cell populations, encoding expectations and prediction error. Crucially, these distinct populations should be reciprocally connected because, algorithmically, every biophysical encoding of an expectation passes messages or signals to an associated prediction error population and vice versa —forward connections convey prediction errors from mechanoreceptors and chemoreceptors (e.g. on apical cells) and backward connections mediate predictions (e.g. from deep vascular cell bundles) —functional asymmetries in forwards (linear) and backwards (nonlinear) connections are mandated by nonlinearities in the generative model encoded by top-down backward connections conveying predictions —vascular cells elaborating predictions (e.g. deep vascular cell bundles) could show distinct (low-pass) dynamics, relative to those encoding error (e.g. cells in root and shoot apices). This follows from the fact that expectations accumulate evidence from prediction errors; thereby suppressing fast (high-frequency) fluctuations in prediction errors —recurrent dynamics are intrinsically stable because they suppress prediction error. In other words, if cells encoding errors excite cells encoding expectations, expectations should inhibit errors—or vice versa
electrophysiology: explains the prevalence of action potentials (APs) and variation potentials (VPs), also known as slow wave potentials (SWPs), in plant electrophysiology [54–58]	—sensory responses are greater for surprising, unpredictable or incoherent stimuli (e.g. sudden changes in salt concentration or mechanical stimulation) —the attenuation of responses encoding prediction error, with perceptual learning. In other words, we would predict that regular fluctuating mechanical, photic or chemical stimulation will entrain plant electrophysiology—and that these induced responses should decay with repetition—and re-emerge with novel stimuli—or, importantly, an omission
psychophysiology: accounts for the behavioural correlates (e.g. growth and phenotypic changes) of physiological phenomena [1]	—predictive processing furnishes a framework in which to model and understand priming and learning phenomena in plants of the sort that underlies omission related responses (see above) and experience dependent plasticity in the way top-down predictions are formed