Friday, October 16
2,050 new papers appeared. We read every one. These 20 are worth your time.
Today's edition
Picked for broad interest, a real step forward and honest claims, with no field allowed to take over.
Dynamics of dopamine transients in human olfactory epithelium during reward learning
Here, we measured sub-second dopamine in the olfactory epithelium of healthy humans during two reward learning tasks with distinct reward types: liquid reward during a conditioning task and monetary reward during an economic exchange game.
AI summary
Researchers measured fast dopamine signals in the nose lining of healthy people during two reward tasks. In both tasks, peak dopamine shifted from rewards to cues that predicted rewards as trials went on. In the money game, this pattern adjusted quickly when reward size or timing changed unexpectedly. These findings fit the known features of brain signals that learn from prediction errors. The nose offers a new, minimally invasive way to record fast dopamine in healthy humans.
Written by Kimi K2.5 from the abstract only. Jev checked every sentence against the authors' words. See where each sentence comes from
PreprintClaims a big stepPopulation structure of human-infecting and environmental isolates of Coccidioides immitis, the fungus causing Valley fever
Most patient isolates derived nearly 100% of their ancestry from a genetic cluster shared with soil isolates, suggesting that the Coccidioides strain inhaled by a human host represents a random draw from the environmental pathogen pool.
AI summary
Coccidioides is a soil fungus that causes Valley fever when people breathe in its spores. The authors sequenced 16 soil samples and 26 patient samples from California's San Joaquin Valley. They found that soil strains show regional patterns, with much genetic mixing between areas. Most patient infections came from one genetic group also found in soil, suggesting random exposure. High genetic diversity exists in soil populations, with neighboring rodent burrows differing by over 13,000 SNPs.
Written by Kimi K2.5 from the abstract only. Jev checked every sentence against the authors' words. See where each sentence comes from
PreprintClaims a big stepReal-world useMonoRace: Robust monocular AI for champion-level drone racing
Here, we present MonoRace, an onboard drone-racing approach that generalizes to a competition environment without any external motion tracking system, using only a monocular rolling-shutter camera and an IMU.
Peer-reviewed journalClaims a big stepReal-world useRapid coral bleaching recovery via symbiont transfer from healthy corals
Here, we demonstrate coral bleaching recovery through contact-mediated transfer of symbionts from healthy corals using small coral fragments.
Peer-reviewed journalClaims a big stepWavelength-selective bidirectional plasticity in self-powered Se/ZnMgO visual synapses
We report here Se/ZnMgO thin film photodiodes in which the photocurrent at zero bias writes and reads the synaptic weight, with a direction of change set by the photon energy alone.
AI summary
The authors made Se/ZnMgO photodiodes that work as visual synapses without needing external voltage. Ultraviolet light (365 nm) strengthens the synaptic weight by 38%, while blue light (455 nm) weakens it by up to 30%. The direction of change depends only on photon energy, determined by defect populations at the interface. A neural network using these devices reached 89% accuracy on handwritten digits, or 85% with noise.
Written by Kimi K2.5 from the abstract only. Jev checked every sentence against the authors' words. See where each sentence comes from
PreprintReal-world useSurvICL: a Foundation Model for in-context Survival Analysis
On the 63 datasets held out from all design decisions, SurvICL outperforms 20 tree-based, deep, and foundation-model baselines without any tuning: it wins 82.9% of pairwise comparisons in integrated Brier score, against 71.3% for the runner-up, the Cox model, and is the first model to rank above Cox on small datasets.
AI summary
Survival analysis predicts when an event will happen from incomplete data, often in small medical studies. The authors created SurvICL, a foundation model that predicts survival curves in-context. SurvICL was pretrained only on synthetic data from a new flexible prior that beat other priors in tests. On 63 test datasets, SurvICL beat 20 other methods without any tuning, winning 82.9% of comparisons. It is the first model to rank above the classic Cox model on small datasets.
Written by Kimi K2.5 from the abstract only. Jev checked every sentence against the authors' words. See where each sentence comes from
PreprintClaims a big stepReal-world useTasting from memory: reconstructing taste from visual cues in the human brain
Taste memory is not simply a re-creation of sensory experience, but a distinct representation, outside the taste system, yet specific enough to guide what we choose to eat.
AI summary
The researchers scanned people's brains while they learned and then recalled links between tastes and pictures. The taste cortex was active during recall and could tell different remembered tastes apart. But it did not use the same brain patterns from when people actually tasted. The authors conclude that taste memory is not a simple replay of tasting, but a separate brain representation.
Written by Kimi K2.5 from the abstract only. Jev checked every sentence against the authors' words. See where each sentence comes from
PreprintCortex-wide single-cell activity mapping in freely moving mice
This enables cortex-wide calcium imaging at single-cell resolution in freely moving mice.
AI summary
The authors built a tiny 3.75-gram head-mounted microscope called COMET to image brain cells in freely moving mice. Tests showed its results match those from standard two-photon microscopes in both neuron detection and calcium signals. The team used COMET to study social interaction and fear learning, finding changes in brain region activity and connectivity before learned responses. The tool offers an open framework for connecting single-neuron activity to whole-brain networks during natural behavior.
Written by Kimi K2.5 from the abstract only. Jev checked every sentence against the authors' words. See where each sentence comes from
PreprintClaims a big stepBioinspired supramolecular thermistor for resolving subtle temperature variations
This work establishes molecularly gated supramolecular thermal transduction as a strategy for ultrahigh-resolution cryogenic temperature sensing.
Peer-reviewed journalReal-world useUnlocking the Tertiary Nitrogen Effect in 3D Anisotropic Covalent Organic Frameworks to Achieve Record Iodine Capture under Industrial Conditions
ECUT-COF-100 was found to give an ultrahigh iodine uptake of 1.51 g/g under simulated industrial conditions (150 °C and 150 ppmv I2), 5.2-fold that of ECUT-COF-99 (0.29 g/g).
Peer-reviewed journalClaims a big stepReal-world usePreclinical and first-in-human evaluation of VG-3927/SAR448851, an oral small-molecule TREM2 agonist for Alzheimer’s disease
We report VG-3927, an oral, CNS-penetrant small-molecule TREM2 agonist that engages the transmembrane domains of TREM2 and its adaptor DAP12 to promote higher-order receptor complex formation and downstream signaling.
Peer-reviewed journalClaims a big stepReal-world useCryptosporidium sex employs a complex male fertilization apparatus that is required for continued infection
Here we identify an elaborate fertilization apparatus and numerous of its molecular components in male gametes.
AI summary
Cryptosporidium parasites must reproduce sexually to keep infections going in the gut. Researchers found a complex fertilization structure in male gametes with two parts. One part touches host cells containing female parasites, and another holds a fusion protein called HAP2. Removing HAP2 stopped fertilization and cured infected mice, also protecting them from future infection. These findings point to specific molecules that could be targeted for new treatments.
Written by Kimi K2.5 from the abstract only. Jev checked every sentence against the authors' words. See where each sentence comes from
PreprintReal-world useSpectral tuning and signaling of diverse and far red–sensitive rhodopsins in mantis shrimp eyes
Here, we report that 17 rhodopsins, mostly from the midband eye region, are involved in color sensing and functionally diverse signaling via Gq/Gi coupling and describe the underlying mechanisms of spectral tuning.
Peer-reviewed journalOne-way electron mirrors from single-layer stacking shifts in rhombohedral graphite
Here we show that single-layer stacking shifts in six-layer rhombohedral graphite provide this platform, acting as one-way mirrors for electrons.
AI summary
The authors found that shifting single layers in six-layer rhombohedral graphite creates one-way mirrors for electrons. In microscopy images, interference patterns appear only on one side of the wall, while the other side stays transparent. Computer models traced this one-way behavior to mismatched layer polarization across the wall.
Written by Kimi K2.5 from the abstract only. Jev checked every sentence against the authors' words. See where each sentence comes from
PreprintClaims a big stepComprehensive characterization of transposable element expression reveals regulatory circuitry and clinical implications in cancers
By integrating multiomics data from ∼10,000 samples and 1000 cancer cell lines, we established an expression atlas of 95,512 individual TEs across 980 subfamilies.
Peer-reviewed journalReal-world useProduct Costing in Data Pipelines
To solve MCDP, we propose CostApp, a hybrid algorithm that achieves both exactness and scalability, even for cyclic and massive-scale pipelines.
AI summary
Large companies spend millions monthly running data pipelines. Assigning these costs to individual products is harder than in physical manufacturing. The authors formalize this as the manufacturing cost of data products (MCDP) problem. They propose CostApp, a hybrid algorithm that handles cycles and massive scale while staying exact. CostApp processed over 28 million entities in under an hour in experiments.
Written by Kimi K2.5 from the abstract only. Jev checked every sentence against the authors' words. See where each sentence comes from
PreprintClaims a big stepReal-world useDense4D: Pixel-Level 4D Human Correspondence
We introduce Dense4D, to our knowledge the first feed-forward model dedicated to dense 4D correspondence across the body, clothing and hair from a set of monocular RGB images, without a body template.
AI summary
It predicts Human Point Maps that give each pixel the 3D location of its surface point at one fixed moment. These maps can locate body points even when they are hidden in some frames but visible in others. Matching coordinates across frames creates 3D tracks without needing camera positions or scene motion estimates. Tests on real people show more accurate tracking and better surface alignment than comparison methods.
Written by Kimi K2.5 from the abstract only. Jev checked every sentence against the authors' words. See where each sentence comes from
PreprintClaims a big stepEmergent coherent-incoherent phonon crossover enables ultralow lattice thermal conductivity in 3D double perovskite
Coherent contributions account for ∼50% of κ l near room temperature and exceeding incoherent contributions above ∼200 K, revealing an emergent coherent-incoherent phonon crossover in a fully crystalline 3D solid.
Peer-reviewed journalClaims a big stepAsymmetric Elongation of Uranyl within a Silver Nanocluster
This work establishes nanoconfinement as a powerful new route to reshape the electronic structure and property of actinide-based materials.
Peer-reviewed journalClaims a big stepThe network structure of the urban revolution
These results demonstrate that decentralized interaction processes, supported in part by mobile intermediary communities, played a central role in sustaining early urban connectivity.
Peer-reviewed journal