With three to five training repeats per direction, separate date curves raised calcium test error by 10.1% in 1,430 retained spine records from 13 mice
Does a spine that stays present keep its visual tuning shape? We tested 1,430 retained spine and viewing-condition interval records from 22 cells in 13 mice. The data came from developing mouse visual cortex. We compared a shared centered curve with free nonnegative gains and means against separate curves for two dates. Each split used five of ten repeat indices for training. Masks left three to five accepted trials per direction in each half. All 252 balanced splits were used. Separate curves raised equal-mouse mean squared test error by 10.14% (95% mouse interval 8.92 to 11.61%). Error rose in every mouse. An added partial-pooling test gave a 0.061% decrease, with an interval that included zero. An added independent-repeat product test gave a positive mean score for nonproportional expected tuning. Its interpretation requires independent errors between repeat halves and dates. Thus, evidence for shape change can coexist with poor prediction by separate noisy estimates. Shared date errors can imitate the shape-change signal. The package preserves the compact data, source checks, code, figures, model scores, and repeat instructions. Calcium responses are not measured synaptic currents.
dendritic spine · visual cortex · structure and function · tuning stability · calcium imaging · repeat validation · partial pooling · measurement error
References
6- usesdoi:10.1038/s41467-025-60825-y
Source study of tracked spine structure and visual calcium responses in developing mouse cortex.
- usesdoi:10.6084/m9.figshare.28738592.v1
Version-specific source of the tracked spine MATLAB table, file 53453786.
- usesIn 10 mice, equal mouse weights reduced the early binocular response correlation rise from 0.051 to 0.007 in 81 fixed spine pairs sci:sha256:c97d0624765e841d3140bb78b82cd769c0456156be9391a76acdff689c7019f1
Its checked compact response extract was used during development. The final input was also rebuilt directly from the original source.
- supportsLarger initial spines had more visual calcium tuning energy but no clear five-day stability gain in 13 mice, assuming independent repeat errors sci:sha256:98a23f49a31f4b6f6d47f561eb9cda620d2301ef1a837d31eb19381522c14dd2
Prior study linked initial optical area to tuning energy and tested later stability. We separate gain and mean changes from tuning shape.
- supportsWith equal spine weights, retained spines gave more summed binocular calcium tuning change than turnover across 13 mice sci:sha256:5c65c354b0a06c8fbb5d91fdb2bf9fef2cf755bacfc0a5b5bd06a770cbb9d6d7
Prior study compared turnover with response change in retained spines. Our test asks whether separate tuning shapes improve repeat prediction.
- usesdoi:10.1016/j.neuroimage.2015.12.012
Independent-repeat products are an established way to remove positive noise bias from response comparisons. We apply this principle to a two-curve Gram determinant.
Cited by
1 publicationPublic catalog entries on Sciscape.
- A fixed ridge test gave a 0.22% calcium prediction gain from spine growth in 13 mice, and removal of one mouse changed the gain's sign
This prior study tested shared and separate shapes with repeats from both dates. We instead predict the later curve using no final response from the test mouse.