---
identifier: iit:result/comolatti-2025-time-grid
title: The directed seven-unit grid of the temporal-flow account specifies 79 distinctions
vetted_by: "Matteo Grasso"
formalism: iit-4.0-2026
summary: >-
  The directed 1D grid that Comolatti, Grasso & Tononi (2025) propose as a
  substrate of temporal flow, in state abcDefG, specifies 79 distinctions (28
  with contiguous mechanisms) and 2651 2-relations (304 among the contiguous
  distinctions), matching the paper's Figure 3B. Each contiguous
  distinction's cause lies toward the "now" and its effect toward the "then".
  Computed with PyPhi, with wedge tripartitions of mechanisms as in the
  authors' notebook.
substrate: >-
  Seven binary units abcdefg in a line. Each unit has a self-connection
  (weight 0.3), a strong lateral connection (0.6) to its neighbour toward a
  (the "then"), and a weak lateral connection (0.1) to its neighbour toward g
  (the "now", where the input array enters). Each unit endorses the grid's
  current state: an input is excitatory when its unit and the target agree in
  that state and inhibitory otherwise, through a logistic with determinism 1
  and no threshold. The input array lies outside the system and matches the
  state, so its drive is non-selective and drops out of the TPM.
state: "(0, 0, 0, 1, 0, 0, 1) — abcDefG, units D and G ON"
phi: 0.00267056407521478
mip: >-
  The directed bipartition into {a, b, c, D} and {e, f, G}, cutting the
  connections from abcD to efG. Of those, only d → e (weight 0.1) is present,
  so the minimum partition severs the weak connection toward the now.
pyphi_version: 2.0.0rc4
script: comolatti-2025-time-grid.py
see_also:
  - iit:measure/phi-s
sources:
  - ref: iit:ref/comolatti-2025
    at: >-
      STAR Methods, "Causal model of the substrate: A directed 1D grid of
      binary units"; Figure 3
---

Comolatti, Grasso & Tononi (2025) propose directed grids as the substrate of
the experience of time flowing. The cause-effect structure this grid specifies
is the one the paper unfolds in Figure 3B. The paper treats the grid as part
of a larger complex, so its claims concern the structure's distinctions and
relations, not the grid's own φ_s.

| quantity | paper, Fig 3B | computed |
|---|---|---|
| distinctions | 79 | 79 |
| distinctions with contiguous mechanisms | 28 | 28 |
| 2-relations among contiguous distinctions | 304 | 304 |
| 2-relations, all distinctions | 2651 | 2651 |

A 2-relation here is either a relation binding two distinctions or a
self-relation, in which one distinction's cause and effect purviews overlap
congruently. The computed counts are 2585 relations between two distinctions
plus 66 self-relations, of which 289 and 15 involve only contiguous
distinctions.

The distinctions are directed. Every contiguous distinction has a cause
leaning toward the now and an effect leaning toward the then:

| mechanism | cause | effect | φ_d |
|---|---|---|---|
| b | c | a | 0.23224 |
| bc | c | b | 0.10332 |
| cDe | De | cD | 0.09689 |
| cDef | Def | cDe | 0.06359 |
| abcDefG | bcDef | abcDef | 0.03380 |

The cause and effect purviews of bc and cDef match the paper's Figure 4. The
script prints all 28 contiguous distinctions.

The authors' notebook evaluates mechanism integrated information over wedge
tripartitions of each mechanism (PyPhi's `WEDGE_TRIPARTITION` scheme), not
over all joint partitions; the paper does not state the scheme. The script
sets that scheme and otherwise uses the default configuration of the IIT 4.0
(2026) formulation. The units have no threshold, as in the paper's printed
activation function (STAR Methods). The notebook encodes the mirror image
of Figure 3 (strong connections toward g, state AbcDefg); mirroring relabels
the units and leaves every quantity unchanged.

To reproduce, run the script with uv, which installs PyPhi 2.0.0rc4 for the run:

```
uv run --with "pyphi==2.0.0rc4" python canon/comolatti-2025-time-grid.py
```
