Storyboard

Active exploration is the deliberate selection of sensory actions to maximize information obtained from the environment. Unlike passive perception, the organism controls what signals it receives: head movements (sniffing, eye saccades, auditory scanning), emission of active signals (echolocation, electrolocation) or spatial displacements.

The information gain IG(a) = H(S) - E[H(S|o_a)] measures how much the action reduces uncertainty about state S. The optimal exploration strategy chooses the action that maximizes IG: this is the principle of 'epistemic curiosity' or information seeking, which accounts for exploratory behaviors observed in mammals and insects.

The exploration-exploitation dilemma is central to adaptive behavior: use current knowledge to maximize reward (exploitation) or explore unknown options that might be better (exploration)? The UCB algorithm solves this in a theoretically optimal way: it adds an exploration bonus (ln(t)/n(a)) that decreases as a stock is visited more, balancing both strategies.

Active sniffing dramatically increases the speed of odorant detection: the exponential model C_recep(t) = C_amb·(1-e^(-v·A·t/V)) shows that higher flow v_sniff fills the nasal cavity faster with the ambient concentration. Rodents sniff at 412 Hz, synchronizing olfactory activity with respiratory rhythm to optimize information acquisition.

Active electrolocation (weak electric fish) disturbs one's own electric field when an object with conductivity other than water enters it: E (_obj - _water)·V_obj/r³. The magnitude and distribution of E in the skin of the fish allows estimating size, shape, distance and conductivity of the object. Gymnotiform fish generate stereotyped exploratory movements to maximize electrolocation information.

>Model

ID:('ky', 589)

Description

ID:(0, 589)