J. Malo1, M.J. Luque1, A. Díez-Ajenjo1, M.C. García-Domene2

1Universitat de Valencia (SPAIN)
2Fundación Oftalmológica del Mediterraneo (SPAIN)
Introduction: learning-by-playing in visual neuroscience:
The students of Visual Neuroscience have diverse backgrounds and often are not familiar with quantitative techniques. As a result, professors have problems to convey experimental and theoretical concepts that are mathematical in nature. In this series of papers devoted to the visual neuroscience class we present a number of computational tools to help the students to learn-by-playing, as opposed to the classical analytical approach in physical sciences. A possibility would be providing the students with closed (stand-alone) tools that can be manipulated to illustrate the concepts [1]. There is a bigger-risk bigger-gain didactic option: here (and in [2]) we provide the students with open tools that can be combined in friendly programming environments such as Matlab/Octave [3].

In this work we use our experience in motion perception [4] to present didactic tools for the simulation of the measurement of receptive fields and tuning properties of motion sensitive cells. Using linear models of V1 and MT cells [5] as unknown black-boxes, the students determine the tuning band of such systems by recording the responses they give to moving gratings or noise.

The proposed virtual lab: characterization of virtual V1 and MT neurons:
Our aim is that students become familiar with the concepts of spatio-temporal frequency and speed tuning. We provide them with software with which they can reproduce the different stages they would follow with real cells in a physiology laboratory, where they would first record the responses of the target cell to stimuli specially chosen and analyze these responses [5]. To this end we need (1) software to generate the stimuli, (2) software to define the neurons, and (3) software to compute the response(s) given the stimuli and the sensor(s). In this work we present such tools. Therefore, the students focus on the two points of interest: decide which are the appropriate stimuli and analyze the receptive fields and tuning curves obtained. The difficulties of calculus are smoothed away.

In this paper we show typical results obtained when “measuring” the receptive-fields in the Fourier domain and the speed-tuning curves in V1 and MT and present the take home message for the students. The general code for this virtual lab is available at in the section VirtualNeuroTools. The specific didactic exercise with a step-by-step explanation is tuning_experiment.m

[1] M.J. Luque, D. de Fez, M. Carmen García and V. Moncho. Tools for generating customized visual stimuli in visual perception labs using computer controlled monitors. Proc. ICERI 2013 Conf. 2013, pp 6200-6207.
[2] J. Malo, M.J. Luque, A. Díez and M.C García. Matlab/Octave Tools for the Visual Neuroscience Class II: Understanding the excitation patterns in V1 and MT Visual Areas. Submitted to Proc. ICERI 2014 Conf. 2014.
[3] J. Malo. Proyecto Docente e Investigador en Ciencias de la Visión. Universitat de València. 2002. Available on-line at:
[4] J. Malo, J.Gutierrez, I.Epifanio, F.Ferri, J.M.Artigas Perceptual feed-back in multigrid motion estimation using an improved DCT quantization. IEEE Trans. Im. Proc. Vol. 10, 10, pp. 1411-1427 (2001)
[5] E.P. Simoncelli & D. Heeger. A model of neuronal responses in visual area MT. Vis. Res. 38(5): 743-761, 1998.