The afternoon of November 11, 2026

From 15:34 to 16:18 on November 11, 2026, we recorded Mio’s internal state and behavior every two seconds. During that time she looked around at things in her room, sat for a while, and drank water once.

Thirst over time (15:34–16:18), from the record of Mio’s internal state.

A thirst value and a single action

At the start of the record, her thirst value was 6.3%. It rose slowly and steadily, reaching 11.9% at 16:09:08.

At 16:09:00 she began walking toward a cup of water, and 14 seconds later she began to drink. By 16:09:24 the value had fallen to 0.9%.

State, choice, action, experience

Nervroid is designed so that bodily state informs the selection of action. As thirst rises, drinking water becomes a stronger option; drinking brings thirst back down. This record is one occasion on which that loop ran end to end in the actual prototype.

Thirty minutes of behavior, 15:48–16:18: sitting, looking around the room, drinking water.

What this record does and does not show

This is an observation of one scene on one afternoon. The thirst level at which the action occurs varies with the situation, and the record does not show that the same sequence happens reliably every time.

We could record the temporal link between the change in value and the action. Whether Mio “feels” thirst cannot be known from this record. Nervroid is not at a stage where we claim to have confirmed consciousness or subjective experience.

Next, we will examine the link between state and action through longer records and through comparisons under varied conditions.

Talk to us about this research