Spatiotemporal kernel of a three-component differential equation model with self-control mechanism in vision (Q6186788)

This paper analyzes a differential equation model with a self-control mechanism to explain the processing of visual information in the brain and retina. More precisely, the authors state that ``...our study poses the following question: Which cellular mechanisms produce the spatiotemporal receptive field structure which consisting of the temporal biphasic function and Mexican-hat function?''. The lateral inhibition mechanism is known to be one of the most important mechanisms in information processing in primitive visual systems. According to the authors of this paper, the model they propose is a slight extension of the model of lateral inhibition proposed by \textit{C. S. Peskin} [Partial differential equations in biology. New York: Courant Institute of Mathematical Sciences (1976; Zbl 0329.35001)] \[ \begin{aligned} \tau_{e}\frac{\partial u_{e}}{\partial t}=-u_{e}+I(t,x),\quad t>0,\ x\in\mathbb{R},\\ u=u_{e}(t,x)-v,\quad t>0,\, x\in\mathbb{R},\\ \frac{\partial v}{\partial t}=\frac{\partial^2v}{\partial x^2}+\beta u-v,\quad t>0, \ x\in\mathbb{R}, \end{aligned} \] where $I$ is the input light pulse to the receptor, $u_{e}$ indicates receptor excitation, $v$ indicates receptor inhibition, $u$ is the receptor response, $\tau_{e}$ and $\beta$ are constants. After an interesting and brief historical review of the mathematical models proposed to explain the visual system, the paper concludes that there are two main models, ``the differential equation model based on cellular interactions'' and ``the convolution model based on receptive field structures''. This has motivated the authors to raise the question highlighted above. In 2018 they presented the below system of equations slightly modifying Peskin's model \[ \tau_{e}\frac{\partial u_{e}}{\partial t}=d_{e}\frac{\partial^2u_{e}}{\partial x^2}+a(-u_{e}+I(t,x)),\quad t>0,\ x\in\mathbb{R}, \tag{3} \] \[ \begin{cases} \tau_{u}\frac{\partial u}{\partial t}=b(-u+u_{e}(t,x))-\gamma v,\quad & t>0,\ x\in\mathbb{R}\\ \tau_{v}\frac{\partial v}{\partial t}=d_{v}\frac{\partial^2v}{\partial x^2}+\beta u-cv,\quad & t>0,\ x\in\mathbb{R} \end{cases}\tag{4} \] where $I$ is the input, $u_{e}$, $u$, and $v$ are unknown functions, and $\tau_{e}$, $d_{e}$, $a$, $\tau_{u}$, $b$, $\gamma$, $\tau_{v}$, $d_{v}$, $\beta$, and $c$ are positive constants. The main contribution of the paper is to show that the solution of (3) and (4) can be written as a convolution integral form with a kernel that depends on $t$, if the input depends only on $t$, and on $t$ and $x$ in general. Section 2 provides preliminary results, complemented in the following section with interesting numerical results. It ends with two appendices, the first on certain details regarding the Peskin model and the second one with a detailed and rigorous statement of solutions to (3) and (4). Reference list is extensive and up-to-date.