Integrated Calendar

Many visual tasks, such as separation of figures from ground and navigation, benefit from the extraction and the usage of local motion signals. Yet, there are many ways in which local motion signals are being represented (mostly based on mathematical and computational considerations). I’ll begin this talk by presenting a computational work that explored whether specific kinds of local motion signals occur in the natural world (Nitzany&Victor, 2014, Journal of Vision).
Next, I will present the results of a neurophysiological experiment where we recorded from the main visual brain areas of two visually accomplished, but very different, animals—macaque monkeys and dragonflies. We found similar responses to local motion signals across species, which may serve as neurophysiologic evidence that mammalian visual cortex and the visual centers of the dragonfly brain process motion using similar algorithms and may have converged on a common computational scheme for detecting visual motion.
Finally, I’ll present our current work, which extends and manipulates a few machine learning techniques to generate novel stimuli, where specific characteristics, with regards to local motion signals, are being preserved.
If time permits, I will discuss another line of work (Menda et. al., 2014, Current Biology, Shamble et. al., 2016, Current Biology), where we were able to record from neurons of jumping spiders. I will explain our approach that enables us to record from those tiny marvelous creatures and review our main findings with regards to visual and auditory cues.

Abstract: The PNN is a specialized form of extracellular matrix, initially deposited around selected neurons during critical periods of development in specific parts of the brain, interrupted by holes where synapses occur. We postulate that the PNN comprises a longer-lived structural template and that new memories are created by cutting new holes in the PNN or by expanding existing holes to enable formation of new synapses or to strengthen existing ones. A basic premise of this hypothesis is that the PNN, should undergo very low metabolic renewal from the first age at which memories are retained until senescence, whereas the active constituents of synapses turn over much more frequently and would therefore be poorer substrates for permanent information storage, unless they are equipped with incredibly accurate copying mechanisms (R.Y.Tsien PNAS 2013). Experimental tests of the hypothesis:
1.PNN longevity; using 15N Spirulina diet for Stable Isotope Labeling in Mammals (SILAM) we compare the lifetimes of PNN proteins vs. synaptic components in Enriched Environment (EE) vs. Conventional Cages (CC), ending the pulse-chase by changing to 14N diet at P45. Analysis by Multidimensional Protein Identification Technology (MudPIT) of four different brain areas indicate:
a. Low turnover rate for PNN proteins while synaptic proteins were at the noise level of 15N /14N ratio.
b. Higher turnover of PNN proteins in EE vs. CC cages
c.Variability in the retention of 15N in PNN proteins between brain areas.
2.Localization of the long-lasting proteins; Imaging of 15N /14N ratio using Nanoscale secondary ion mass spectrometry (nanoSIMS) localized and verified the MudPit finding that PNN turnover is very slow.
3. Spatial occupation of the PNN holes; 2 dimension electron microscopy (EM) and 3D volumes of Serial Block Face Scanning EM reveal that neurons engulfed in PNN have more than 95% of their plasma membrane surface occupied by PNN or synapses.
4. Inhibition of PNN holes modulation during strong memories acquisition; we examined the role and timing of matrix metalloproteinases (MMP) activity in memory consolidation using pharmacological inhibitors in a fear-conditioning paradigm. Our results demonstrate that MMP inhibition during fear induction:
a. Does not affect acquisition
b. Significantly impairs long-term memory (30 days)
c. Is dose dependent
d. That memory impairment increases with time.

So far the hypothesis is supported by the results of the above tests.

Photonics is the only platform for quantum information processing with the potential to operate at room temperature, in ambient conditions, without the need for cryogenics, high vacuum or electromagnetic shielding. But it cannot be scaled up because logical operations in linear optics are fundamentally non-deterministic. My research has focussed on a route to scalable photonics by actively synchronising successful operations with quantum memories — devices that can store and release photons on-demand. In this talk I will review our approach to this challenge and present our most recent results demonstrating noise-free storage of GHz-bandwidth heralded single photons in warm vapour.