THEORY-1321: Froese proposes that irruptions combined with attunement, modelled on Watson et al.'s self-optimization of a Hopfield network by occasional resets to arbitrary states plus Hebbian learning, answer the libertarian's randomness challenge, how behaviour can be effective though underdetermined; but in the model offered the resets are arbitrary, so what makes the behaviour effective is the learned attractor landscape, and nothing in it distinguishes a motivated irruption from noise

Source

Froese (2023), LIT-1950, §§4, 5.2.3 and 6, with Fig. 4 (from Froese, Weber, Shpurov and Ikegami); read in NOTE-1521. The self-optimization model is Watson, Buckley and Mills (2011), which the record does not hold.

What was actually shown

The challenge (§4). A libertarian account must reject “the complete material determination of agent behavior” while “avoiding falling into the opposite extreme of completely random agent behavior”.

The proposal (§5.2.3). In the self-optimization model, a Hopfield network’s states are “occasionally ‘reset’ to an arbitrary configuration” and allowed to relax, with Hebbian learning; over time the network “generalizes over the set of visited equilibria” and reaches deeper ones. Froese proposes that motivations, expressed as irruptions, “could serve the role of the ‘resets’”, and that this “holds potential to provide a novel systems theoretic answer to the main concern raised against libertarian interpretations … how an agent’s behavior can be effective even when it is underdetermined”.

What the model, as described, does. The improvement comes from the resets’ arbitrariness plus learning. Nothing in the description makes it depend on the resets coming from motivation rather than from any other source of arbitrary perturbation, and by Axiom 2 no measurement could tell the two apart. So the model shows that random perturbation plus plasticity can make behaviour effective. Whether motivated perturbation adds anything is what would have to be shown.

What this does not say

• That Irruption Theory is wrong. A version in which the timing or target of resets carries the motivation would be different; the paper does not give one.
• That self-optimization is uninteresting. It is a case where noise helps an attuned system, which supports Froese’s point that irruptions “do not stand in contrast to adaptive behavior” (abstract).
• That the luck objection is answered or unanswered in general. Only that this proposal places effectiveness in attunement, an agent’s history, which is not the indeterminacy.