The necessity of both signs and dynamics in describing informational processes in living systems have been advocated at least since the 60s of the last century (Polanyi, 1968; Pattee, 1969). A useful conceptualization of their relation seems to be that of control, where informational structures are understood as functional constraints on relevant degrees of freedom on various time-scales (Pattee, 1973; Rączaszek-Leonardi & Kelso, 2008). This move liberates theories of communication, including language, from the cumbersome „coding metaphor”, which hinders at least some aspects of constructing a viable semantic theory (Brette, in press).

The same move, however, calls for new frameworks for capturing the informational processes and in my talk I will propose that such a framework may consist in integrating dynamical systems analysis with the semiotic approach (Rączaszek-Leonardi et al., 2018). Semiotic approach helps dynamical systems theory recognize various types of constraints, while dynamical systems approach brings useful operationalization of the key notion of “interpretation”, which to many may seem esoteric and to some overly internalist.

This theoretical background allows for rethinking the symbol emergence process: rather than “simply” seeking symbol grounding (often in the form of mapping) it forces a researcher to recognize a complex semiotic infrastructure on which symbolic control of dynamical processes constitutively depends (Deacon, 1997; 2011). Useful reformulations of the aspects of abstractness and compositionality follow. In my talk I will focus on the relation between the complexity of control and compositionality in mother-infant interactions in development and show how empirical research on interactions over the ontogenetic time scale may inform computational models of the general principles of symbol emergence.