Pollination networks are commonly analysed as sets of realized plant–pollinator links, whereas the behavioural processes producing those links occur through sequential decisions by individual foragers. This scale mismatch makes it difficult to determine when network structure reflects persistent individual choice, changing resource opportunity, or aggregation across heterogeneous insects. This article develops an ecological framework that separates floral valuation, learning, constancy, switching, exploration, repeated visitation, pairwise link realization, and network emergence. These processes are treated as analytically distinct rather than as interchangeable descriptions of preference. The central proposal is that individual floral decisions contribute to network structure only through repeated realization within an opportunity set constrained by floral availability, phenology, morphology, competition, and observation scale. Consequently, an observed network edge is not equivalent to a cognitive preference, and network rewiring is not direct evidence of individual switching. The framework predicts that network-level interpretation should improve when behavioural persistence is measured across explicitly defined temporal windows and individual trajectories are nested within repeated community sampling. It identifies key validation needs, including separation of visitation from pollen transport, individual from colony-level inference, and behavioural updating from species or resource turnover.
