Where does heritability show up in twins watching the same film?

Three measures of the same stimulus

In the Human Connectome Project's 7-tesla release, 178 people passively watched short film clips across two sessions; the sample holds 51 monozygotic and 34 dizygotic twin pairs. Heritability of the raw BOLD time course came out at 0.064 on day one and 0.068 on day two. Heritability of the functional connectivity profile, computed from the same scans, came out at 0.36 on day one and 0.37 on day two. The stimulus is the same, the scan is the same, the people are the same; the only thing that changes is what gets measured.^[1]

The third measure is the neural timescale, which describes how long a window a region uses to respond to the stimulus. Its global heritability is 0.38 on day one (p = 0.003) and 0.75 on day two. That jump is a warning in itself: an estimate that almost doubles between two sessions does not yet count as a settled measure. The same authors report a steadier value at the level of individual regions; the volatility in the global figure is consistent with a small sample and the noise of a single session.^[1]

What alignment takes away

The most instructive part of the study recomputes heritability after applying alignment methods that superimpose individual functional maps. Raw BOLD heritability drops by 33 percent on day one (95 percent confidence interval 25–40) and connectivity profile heritability by 39 percent (32–46). A substantial share of the apparent heritability comes from where a response lands on a person's cortex rather than from the response the tissue gives to the stimulus.^[1]

There is a finding on the other side too: measured during films, the connectivity profile is 20 percent more heritable than at rest on day one and 30 percent more on day two for sensory networks. The stimulus, in other words, appears to make heritable structure visible. A plainer account could produce the same result: a film generates a stronger and more regular signal than rest, and a measure with less noise returns a higher heritability estimate.^[1]

When does a measure become a candidate marker?

For an imaging measure to be used as a trait carried in families, two conditions have to hold: it must be heritable, and it must give the same value when the same person is measured again. The connectivity profile meets the first condition far better than the raw signal, and by staying at 0.36 and 0.37 across two days it comes close to the second. The neural timescale meets the first condition strongly while swinging from 0.38 to 0.75 on the second. No measure here is usable in the clinic today, but which measure is worth working on has become clearer.^[1]

Who was measured is part of the result as well. Every participant is between 22 and 36 years old, and close to 90 percent identified as White. Heritability describes a ratio that depends on the genetic and environmental variety of the group it is measured in, so these numbers belong to a narrow group. Whether the same measures would fall in the same order in a childhood or an elderly sample cannot be drawn from this study.^[1]