A condition that preserves the symmetry of a system and restricts its degrees of freedom. When the system is perturbed, it may lead to symmetry breaking. Applied to behavior, a constraint is a boundary condition that eliminates or restrains certain configurations of action while permitting or enabling others. There are two classes of constraints: holonomic or law-governed constraints that restrict without having any material embodiment in the system (e.g., Newton’s laws of motion) and non-holonomic or rule-governed constraints that are physically embodied in the system (e.g., a schema) and which serve as prescriptions for action. Computational models incorporate the latter constraints (e.g., model architecture, training algorithm, training regime). In terms of cognition, a constraint is anything that makes representational development selective in what becomes represented or how the representation occurs. Typically, a constraint has the function of guiding the child’s learning within biological boundaries to the extent to which the child can process information and prepare action. In developmental biology, four classes of holonomic constraints are recognised: developmental constraints, physical constraints, morphological constraints and phyletic constraints.
See Adaptation, Canalization, Computational models, Control parameter, Core concepts, Degrees of freedom (or Bernstein’s) problem, Developmental biology, Embodiment, Foundational knowledge, Information-processing theories, Laws of nature, Life course analysis, Newton’s laws of motion, Phylogenetic gradualism and punctuated equilibrium, Schema, Self-organization, Symmetry breaking (and perseveration)