Anticipatory behaviour and predictive coding in plants

Child (1924) framed plant behaviour as developmental response. He argued that many developmental processes stand in direct relation to environmental factors, inside or outside the organism. When those relations are altered, for example by isolating parts, the parts respond with altered development¹. He concluded that by far the greater part of what biologists call plant behaviour consists of such changes in development, growth and differentiation¹. This framing treats plant behaviour as an adjustment of growth to conditions.

Calvo and Baluška (2015) took plant anticipatory behaviour as their case. Their working hypothesis was that adaptive plant behaviour needs a mechanism that predicts sensory states². They separated weaker readings, in which anticipation relies on an internal model of environmental sources of stimulation, from stronger forms that do without internal modelling. They held that the stronger forms could not be discarded beforehand³. Lyon (2015) noted that recent studies of anticipatory behaviour in bacteria involve responses to predictable stressors⁴. Bruineberg, Kiverstein and Rietveld (2018) described predictive coding for the brain as top-down predictions matched against sensory information, producing prediction errors⁸. They said it should not be conflated with the free-energy principle⁸.

Calvo, Baluška and Sims (2016) framed a direct comparison of feature detection and predictive coding models of plant behaviour. They did not claim that anticipation is sufficient for intelligence, since basic predictive abilities exist in artificial neural networks⁵. They claimed it is necessary for minimal intelligence⁵. On that basis, confirming the feature detection model would falsify the minimal intelligence hypothesis. Otherwise the conjecture would be kept provisionally⁶. They also said predictive coding is not the only alternative to feature detection. An ecological, non-model-based account might explain anticipation through the organism-environment coupling itself. Whether that is plausible they left open⁷.

Khattar and colleagues (2022) surveyed interdisciplinary views of plant intelligence. They argued that most disagreements come from semantic issues rather than from the behaviours themselves. Some scientists take metaphors literally and others take literal terms as metaphors⁹. They also reported that most respondents did not notice the difference between fixed, genetically determined behaviours and intelligent ones⁹.