Search bioRxiv⌕ Search

Biology subjects

Karadas, M.

Publications and source records attributed to Karadas, M..

2 recordsLinked to original sources

Temporal sequence geometry enables odor recognition and generalization

Neural activity sequences are observed throughout the brain, yet their computational roles remain elusive. In mammalian olfaction, olfactory bulb mitral and tufted cells (MTCs) encode odors with precisely timed activity patterns that tile the respiration cycle. While animals can identify odors independently of concentration within the first 100 milliseconds of inhalation, the structure governing these sequences and the role of activity extending beyond this early window remains unclear. Here, using 2-photon calcium imaging with sub-sniff resolution, we show that odor-evoked MTC sequences propagate as wavefronts through a low-dimensional odor tuning space, where timing is predicted by the tuning similarity between neurons rather than physical location. While early portions of these sequences are concentration-invariant, providing a stable anchor for odor identity, later portions systematically co-activate similarly tuned MTCs across odors, tracing the geometry of the tuning manifold. We propose a role for this later sequential activity in training the piriform cortex to learn perceptually generalizable odor representations. Using a model of Hebbian learning through sequences (HeLSeq), we demonstrate that sequential activity can reinforce synaptic connections from similarly tuned MTCs onto common piriform cortical neurons, enabling rapid generalization to novel odors from the earliest moments of inhalation. These findings support a geometric view of activity sequences and establish a general principle by which temporal sequences scaffold unsupervised manifold learning between brain networks.

neuroscience↗

From labels to latents: revealing state-dependent hippocampal computations with Jump Latent Variable Model

Neural activity is usually interpreted by imposing external labels (e.g., stimuli or position during locomotion) and decoding within that space (e.g. replay). While powerful, such supervision can mask structure in the data that do not correspond to the label. Unsupervised methods, in turn, often assume smooth latent dynamics and miss genuine discontinuities. We introduce a conceptually simple, computationally efficient latent variable model that infers both (i) the latent variables organizing population activity and (ii) whether their dynamics are continuous or fragmented in time. Fitting reduces to an expectation-maximization (EM) procedure that alternates two operations familiar to systems neuroscience--tuning-curve estimation and label decoding--without requiring external labels. Applied to rodent hippocampal spike recordings, the model reveals distinct population patterns at the same physical position that supervised spatial decoding fails to detect. While learned latents exhibit place-field-like tuning, their reactivation patterns are better distinguished by behavioral states. The model further identifies a continuity-fragmentation axis that characterizes population activities across sleep-wake brain states that is modulated by cholinergic inputs. By not relying on externally imposed spatial labels, our approach exposes structure that supervised approaches obscure and provides a powerful tool for datasets lacking behavioral tracking.

neuroscience↗