# Figures of journal note 27 (protocol 024, the two-signal decision rule)

Both figures are drawn by `draw_decisions.py` from the CSVs that `make_decisions_csv.py` wrote from this package's
`results/decisions-a.json`. No other input, no image generator.

## Figure 1: `decision-steps` (`.png`, `.svg`; data `decision-steps.csv`)

**Caption.** The decision variable v of all 336 test steps of the six network instances, by stimulus. Signal A is a
grating at 45°, signal B one at 135°; numbers are contrasts in percent. Each dot is one step; within a stimulus the
six columns of dots are the six instances (pynn_seed 8, 10, 11, 12, 13, 14). Dashed lines: the thresholds at ±0.32,
fixed on the pilot instance beforehand. Above the upper line the rule takes action B, below the lower line action A,
between them it holds. All 336 steps took the intended action. Simulation only.

**Alt text.** Dot chart titled "Every test step fell on the intended side of the thresholds". Vertical axis: decision
variable v from −1.7 to 1.5, with a shaded band between dashed lines at −0.32 and +0.32 labelled hold. Eight stimuli
along the horizontal axis, each with six columns of dots. A 60: dots around −1.15 to −1.6. A 15: around −0.5 to −0.85.
A 30 + B 15: around −0.6 to −1.05. A 15 + B 15 and A 30 + B 30: inside the band, between −0.3 and +0.2. A 15 + B 30:
around +0.5 to +1.0. B 15: around +0.4 to +0.75. B 60: around +0.85 to +1.4.

## Figure 2: `decision-relay` (`.png`, `.svg`; data `decision-relay.csv`)

**Caption.** For each of the four plaids and each of the six instances: the mean decision variable of the circuit's
response to the plaid (violet), and the relay rule's value, the sum of the mean decision variables of the two gratings
shown alone (open circle; filled black where it falls on the wrong side of the thresholds). The relay rule errs in 5
of 24 cases: it holds in four instances where B is the stronger signal, and acts in one instance on an equal plaid.
Reported beside the outcome, with no decision role.

**Alt text.** Dot chart titled "A relay of the single responses would have erred in 5 of 24 cases". Four groups of six
pairs of dots on a vertical axis from −1.1 to 1.1 with a shaded hold band between ±0.32. A 30 + B 15, intended A:
circuit dots at −0.70 to −0.87, relay dots at −0.35 to −0.70, all below the band. A 15 + B 15, intended hold: all
dots inside the band. A 30 + B 30, intended hold: circuit dots inside the band; one relay dot at −0.40, below it,
filled black. A 15 + B 30, intended B: circuit dots at +0.60 to +0.82; relay dots at +0.06 to +0.38, four of them
inside the band and filled black.

Model: 2 mm reduction of LSV1M (Antolík et al. 2024, CC BY 4.0), simulated with Mozaik and NEST 3.4.
