Biology subjects

Warren, K.

Publications and source records attributed to Warren, K..

6 recordsLinked to original sources

bioRxiv2026

Head-level lesion-symptom mapping of picture naming in vision-language models

Researchers in artificial intelligence increasingly intervene on language models to study how their functions are organized, silencing weights and attention components in ways reminiscent of the brain lesions long used to map language in post-stroke aphasia. Under this program of mechanistic interpretability, a common move is to ablate an attention head and read the resulting drop in a behavior as evidence that the head implements it. How much that move can establish about where a behavior is computed remains unclear, because a head whose removal disrupts a behavior is necessary for it but need not be the site where it is computed. We examined this question for the picture naming task, taking the loss of naming (anomia) that defines aphasia as the behavior of interest. Across six vision-language models, spanning three language backbones and a range of parameter scales, we ablated each attention head in turn during single-word picture naming and measured accuracy before and after. The degree of localization varied widely across models. In LLaVA-1.6-Vicuna-13B, a single early head (layer 0, head 20) was necessary: removing it alone reduced naming accuracy from 0.99 to 0.006. The same head was not sufficient, because retaining it while ablating the other 1{,}599 heads also produced 0\% accuracy. Two Mistral-backbone models (LLaVA-Mistral-7B and Idefics2-8B) had no critical head. An early-layer dependence was present in every model but varied in strength, whereas dependence on any single head ranged from dominant to absent. Within Qwen2.5-VL, a dominant head was present in the 7B model but not the 3B model, indicating that this concentration emerged with scale rather than being fixed across a model family, and it was not explained by attention type. These results show that ablating the head whose removal disrupts naming does not establish that the head computes the behavior, that the result generalizes across models, or that the behavior localizes to a head at all.

neuroscience↗

bioRxiv2026

Cephalo-pelvic covariation and sexual dimorphism are disrupted in hybrid mice: implications for the human obstetrical dilemma

Cephalo-pelvic disproportion in humans has traditionally been interpreted through the obstetrical dilemma framework, assuming a trade-off between bipedal locomotion and childbirth. However, cephalo-pelvic covariation and pelvic sexual dimorphism might be common adaptations to parturition among mammals. We use a controlled hybridization model in mice to test whether cephalo-pelvic covariation and pelvic sexual dimorphism are population-specific, genetically structured, and sensitive to hybridization. We analyzed skull-pelvis variation and covariation, as well as sexual dimorphism of pelvic morphology across four divergent wild-derived mouse strains and their hybrids. Hybridization induced consistent cranial and pelvic size enlargement. Females exhibited significant cephalo-pelvic shape covariation, characterized by an association between rounder, wider birth canals and larger neurocrania, consistent with functional integration under obstetric selection. Hybrids showed disrupted size covariation, increased pelvis shape variance, and reduced cephalo-pelvic integration. Pelvic sexual dimorphism was systematically reduced in hybrids. Cephalo-pelvic covariation and pelvic sexual dimorphism are not exclusive to bipedal or encephalized species. They likely reflect widespread selection on birth canal morphology in mammals and have a complex genetic basis sensitive to hybridization. These findings weaken a human-exclusive interpretation of the obstetrical dilemma and highlight genetic introgression as an understudied factor shaping cephalo-pelvic integration and disproportion risk in mammals, including humans.

evolutionary biology↗

bioRxiv2026

Oscillating Hypercapnia Induces Neural Abundant Protein Efflux and Potential Depletion in Health and Chronic Traumatic Brain Injury

Emerging preclinical and clinical evidence suggests that low frequency hemodynamic oscillations drive CSF flow, which in turn mediates glymphatic clearance. The current study investigated whether CO2-induced low frequency hemodynamic oscillations during magnetic resonance imaging would increase clearance of proteins (glial fibrillary acidic protein, neurofilament light chain, ptau217 and brain-derived tau) from brain to blood, and temporarily improve cognitive performance in individuals with chronic traumatic brain injury (TBI) and age/sex-matched healthy controls. Results indicated that cerebrovascular reactivity, normalized CSF volume, and predicted brain age significantly differed between chronic TBI and controls, while bulk CSF flow differed only at trend levels. Multiple protein concentrations were significantly increased at [~]45 minutes post-hypercapnia, decreased at [~]90 minutes, and returned to pre-hypercapnia levels by [~]150 minutes. Protein efflux was more strongly associated with total CSF volume and total white matter volume rather than cerebrovascular reactivity or bulk CSF flow. Both groups exhibited reduced cognitive interference post-hypercapnia, and hypercapnia associated symptoms quickly returned to baseline levels. In conclusion, hypercapnia temporarily increases clearance of multiple neural abundant proteins into blood, and this effect is moderated by atrophy. Current results suggest that hypercapnia may therapeutically combat pathological protein aggregation post-trauma, and prophylactically during normal aging.

neuroscience↗

bioRxiv2023

Evidence for deliberate burial of the dead by Homo naledi

In this study we describe new results of excavations in the Dinaledi Subsystem of the Rising Star cave system, South Africa. In two areas within the Hill Antechamber and the Dinaledi Chamber this work uncovered concentrations of abundant Homo naledi fossils including articulated, matrix-supported skeletal regions consistent with rapid covering by sediment prior to the decomposition of soft tissue. We additionally re-examine the spatial positioning of skeletal material and associated sediments within the Puzzle Box area, from which abundant H. naledi remains representing a minimum of six individuals were recovered in 2013 and 2014. Multiple lines of evidence exclude the hypothesis that skeletal remains from these three areas come from bodies that decomposed on the floor of the chamber or within a shallow depression prior to burial by sediments. The spatial positioning of skeletal material, the topography of the subsystem, and observations on sediments within and surrounding features exclude the hypothesis that rapid burial by sediment was a result of gravity-driven slumping or spontaneous movement of sediments. We present a minimal hypothesis of hominin cultural burial and test the evidence from all three areas, finding that this hypothesis is most compatible with the pattern of evidence. These results suggest that mortuary behavior including cultural burial was part of the repertoire of Homo naledi.

paleontology↗

bioRxiv2023

Pelage Variation and Morphometrics of Closely Related Callithrix Marmoset Species and Their Hybrids

BackgroundHybrids are expected to show greater phenotypic variation than their parental species, yet how hybrid phenotype expression varies with genetic distances in closely-related parental species remains surprisingly understudied. Here we study pelage and morphometric trait variation in anthropogenic hybrids vailable at the end of the article between four species of Brazilian Callithrix marmosets, a relatively recent primate radiation. Marmoset species are distinguishable by pelage phenotype and morphological specializations for eating tree exudates. Here, we (1) describe qualitative phenotypic pelage differences between parental species and hybrids; (2) test whether significant quantitative differences exist between parental and hybrid morphometric phenotypes; and (3) determine which hybrid morphometic traits show heterosis, dysgenesis, trangression, or intermediacy relative to the parental trait. For morphometric traits, we investigated both cranial and post-cranial traits, particularly as most hybrid morphological studies focus on the former instead of the latter. Finally, we estimate mitogenomic distances between marmoset species from previously published data. ResultsMarmoset hybrid facial and overall body pelage variation reflected novel combinations of coloration and patterns present in parental species. In morphometric traits, C. jacchus and C. penicillata were the most similar to each other, while C. aurita was the most distinct, and C. geoffroyi trait measures fell between these other species. Only three traits in C. jacchus x C. penicillata hybrids showed heterosis. We observed heterosis and dysgenesis in several traits of C. penicillata x C. geoffroyi hybrids. Transgressive segregation was observed in hybrids of C. aurita and the other species. These hybrids were also C. aurita-like for a number of traits. Genetic distance was closest between C. jacchus and C. penicillata and farthest between C. aurita and the other species. ConclusionWe attributed significant morphometric differences between marmoset species to variable levels of morphological specialization for exudivory in these species. Our results suggest that intermediate or parental species-like hybrid traits relative to the parental trait values are more likely in crosses between species with relatively lesser genetic distance. More extreme phenotypic variation is more likely in parental species with greater genetic distance, with transgressive traits appearing in hybrids of the most genetically distant parental species. We further suggest that less developmental disturbances can be expected in hybrids of more recently diverged parental species, and that future studies of hybrid phenotypic variation should investigate selective pressures on Callithrix cranial and post-cranial morphological traits.

zoology↗

bioRxiv2018

Low oxygen post conditioning improves stroke-induced cognitive impairment

Post-stroke cognitive impairment has proven to be notoriously difficult to treat. In the current study, we sought to both better understand cellular changes that underpin cognitive deficits and to consider the potential restorative benefits of low oxygen post conditioning (LOPC). We were motivated to use LOPC as an intervention as it is one of the few experimental interventions previously shown to improve cognitive function post-stroke. Experimental stroke was induced by photothrombotic occlusion in adult male C57BL/6 mice. Mice were randomly assigned to either a normal atmospheric air exposure or low oxygen (11% O2) exposure groups three days post-occlusion. On day 17 post-stroke, mice were euthanized for histology or biochemical analyses. Stroked mice exposed to LOPC was associated with marked reductions in amyloid-beta both in its absolute level and in the extent of its oligomerization. Exposure to LOPC post-stroke also improved cellular deficits induced by stroke including an increase in vessel density, a reduction in vascular leakage, and restoration of AQP4 polarisation. Critically, stroked mice exposed to LOPC exhibited robust improvements in cognitive function post-stroke, assessed using a touchscreen based paired-associate learning task. These findings provide compelling pre-clinical evidence of the potential clinical utility of LOPC for enhancing recovery post-stroke.

neuroscience↗