We study the behavior of a minimal model of synaptically sustained persistent activity that consists of two quadratic integrate-and-fire neurons mutually coupled via excitatory synapses. Importantly, each of the neurons is excitable, as opposed to an oscillator; hence when uncoupled it sits at a subthreshold rest state. When the constituent neurons are mutually coupled via sufficiently strong fast excitatory synapses, the system demonstrates bistability between a fixed point (quiescent background state) and a limit cycle (memory state with synaptically driven spiking activity). Previous work showed that this persistent activity can be stopped by an excitatory input that synchronizes the network. Here we analyzed how this persistent state reacts to partial synchronization. We considered three types of progressively more complex excitatory synaptic kernels: delta pulse, square, and exponential. The first two cases were treated analytically, and the latter case numerically. Using phase-plane methods, we characterized the shape of the region, such that all orbits starting within it correspond to infinite spike trains; this constitutes the persistent activity region. In the case of instant coupling, all such active orbits were neutrally stable; in the case of noninstant coupling, the activity region contained a unique stable limit cycle (so the activity region was the basin of attraction for the limit cycle). This limit cycle corresponded to purely antiphase spiking of two neurons. Increasing synchronization shifted the system toward the border of the activity region, eventually terminating spiking activity. We calculated three measures of robustness of the active state: width of the activity region in the phase plane, critical level of synchronization that can be tolerated by the persistent spiking activity, and speed of reconvergence to the limit cycle. Our analysis revealed that the self-sustained activity is more robust to synchronization when each individual neuron is closer to SNIC bifurcation (closer to being an intrinsic oscillator), the recurrent synaptic excitation is stronger, and the synaptic decay is slower, which is in agreement with the existing data on local circuits in the cortex that show sustained activity.
The paper discusses several problems which have been observed during the development of the corpus of West Circassian and proposes that their solutions should involve the possibility of multiple analyses. It is argued that this is related to certain properties of the constructions under discussion which are reflected in variation observed among the speakers of West Circassian as well as in the diachronic instability of these constructions.
Both examples and verbal explanations play an important role in learning new concepts and categories. At the same time, learning from verbal explanations is not accounted for in most category learning models, and is not studied in the traditional category learning paradigm. We propose a rational category communication model that formally describes the process of communicating a category structure using both verbal explanations and visual examples in a pedagogical setting. We build our model based on the assumption that verbal instructions are best suited for communication of crude constraints on a category structure, while exemplars complement it by providing means for finer adjustments. Our empirical study demonstrates that verbal communication is indeed more robust to changes in stimuli dimensionality, but that its efficiency is adversely affected when distinguishing between categories requires perceptual precision. Communicating through examples has a reversed pattern. We hope that both the proposed experimental paradigm and the computational model would facilitate further research into the relative roles of verbal and exemplar communication in category learning.
In my paper, I will analyze a special feature of clausal complements in the Bzhedug dialect of Adyghe, a polysynthetic language of the West Caucasian family. While in many languages, clausal complements cannot trigger verbal agreement, in Adyghe, the matrix verb can bear plural agreement with a clausal complements, along with the default singular agreement. If there is a coordinate structure including several clausal complements, the agreement slot they correspond to can contain a plural marker. However, this is not obligatorily the case. I will discuss the conditions of this unusual agreement pattern. I will show that the possibility of agreement depends on at least two syntactic parameters: namely, the syntactic position of the clausal complement and the morphological verb form which is used in the complement clause. Non-standard agreement is also subject to a significant inter-speaker variation. Sometimes speakers who do not allow agreement with clausal complements admit instead long distance agreement. The most unexpected thing is that even if the complement clauses are marked with a (typically) non-argument suffix, they can control agreement. This points to the fact that neither the canonical view of the pronominal argument hypothesis, nor classical approaches to agreement, represented in many works on European languages is plausible for Adyghe. The data of this language must be accounted for in a perspective that regards verbal personal markers and verbal arguments as types of items, which are not isomorphic to each other and which both play role in the agreement marking.