attention

the focus a single neuron projects onto a target particle or axon. where collective $\phi^*$ is the whole graph's attention distribution, a neuron's attention is its individual contribution to that distribution — how much of its weight lands on each target.

attention is a derived focus quantity, not a stored field. it is computed by the tru from two write paths a neuron uses to place weight:

• will — broad staking. locking $CYB for a duration produces will, auto-distributed across every cyberlink the neuron creates. each link receives a share; longer lock → more will → more attention per link.
• conviction — per-link. the box $(\tau, a)$ a neuron locks into a specific edge directs attention to that target precisely. fine-tuning beyond the broad will default.

aggregation into collective focus

individual attention is the per-neuron term that sums into the effective adjacency the tri-kernel runs on:

$$A^{\text{eff}}_{pq} = \sum_\ell a(\ell)\cdot \kappa(\nu(\ell))\cdot f(m(\ell))$$

each neuron's attention on $(p,q)$ — its will-derived share plus per-link conviction, weighted by its karma $\kappa$ and the ICBS market price $f(m)$ — is one summand. collective $\phi^*$ is the fixed point of the tri-kernel over the sum of all neurons' attention. attention is the input; $\phi^*$ is the convergent output.

reading attention

attention is exposed as a read at the cybergraph boundary via query(from, to) — see cybergraph/specs/attention for the query-interface view. the query mechanism is cybergraph's; the quantity it returns is defined here.

see focus for the collective distribution · will for broad staking · box for per-link conviction · karma for the trust weight · focus-flow for the computation.