Contents lists available at ScienceDirect Biologically Inspired Cognitive Architectures journal homepage: www.elsevier.com/locate/bica Research article Towards a model of visual recognition based on neurosciences Adrían González-Casillas ⁎ , Luis Parra ⁎ , Luis Martin ⁎ , Cynthia Avila-Contreras, Raymundo Ramirez-Pedraza, Natividad Vargas, Juan Luis del Valle-Padilla, Félix Ramos ⁎ Department of Computer Science, Center for Research an Advanced Studies of the National Polytechnic Institute (CINVESTAV IPN) Unidad Guadalajara, Guadalajara, Jalisco, Mexico ARTICLE INFO Keywords: Brain model Visual features Perception Cognitive architecture Bioinspired model Visual memory ABSTRACT Cognitive sciences and computer vision have proposed diverse models to acquire, transform and interpret visual information, mainly aimed to achieve realistic, yet efficient approaches to those capacities. One of the key aspects of visual processing is the identification of objects in the scene, that entails the perceptual association of visual features with semantic information extracted from memory. In this study, we present a model for visual recognition that resembles the way the human’s brain interacts to achieve this process. The model describes the processes in V1 and V2 to extract features of lines, angles, and contours; as well as a template matching process in ITC, that uses early low spatial frequency visual information to bias the available comparisons. Operations of prefrontal areas DLPFC and VLPFC to maintain the representation and OFC to give a response are also described. Our proposal is intended to be the basis to treat visual information in a broader cognitive architecture. We find that matching of ITC templates provide a general and biologically inspired representation for objects. We also show how the use of low spatial frequency visual information can lead to a faster identification process when previous data exists. This is achieved by selecting a small number of ITC templates to handle the incoming bottom-up input. Introduction In optimal conditions, vision is the main source of information from the environment, therefore, it is the most studied sensory system and crucial to understanding human perception. Visual processing involves mechanisms to generate internal abstract representations, by applying multiple transformations to the light of environmental objects that reaches photoreceptors in the eye. Recognition refers to giving a meaning to such representations (Albright, 2015, chap. 28), regardless of simplicity, and it is shaped by the current sensory activations, past sensory experiences and associa- tions between these experiences. Effective and efficient visual recognition is critical in various sce- narios, like detecting dangerous predators hidden in the woods or in- terpreting a red traffic light while driving. Visual recognition plays an important role in setting basic information required to generate plans to interact with the environment, and then be able to make decisions over possible actions to satisfy goals. Russell and Norvig (2009) state some commonly required properties that a general artificial intelligence should include, such as being capable of sensing, perceiving, learning, representing knowledge, and making decisions. The issue is often how all these capabilities coexist in the same schema. Cognitive Architectures (CA) are useful approaches to construct this type of systems, because they aim to describe the struc- ture and interactions of the human mind’s functions, and how to in- tegrate them. The main motivation of this work is to build a model of visual processing for virtual entities that resemble the way humans do and contribute to a better comprehension of the mechanisms and functions involved in the process of visual object recognition tasks. The emphasis is on bottom-up and top-down, as well as the process importance when encompassed in a larger a cognitive system, such as a cognitive archi- tecture. In this paper, we present a cognitive model for visual processing and object recognition that can be integrated with a broader cognitive ar- chitecture, by setting the basis of the different processes and brain areas involved. This model has modules associated with brain areas that perform operations of one or various Cognitive Functions (CF). The CF provide specific human-like capabilities to the overall CA in which these are integrated. https://doi.org/10.1016/j.bica.2018.07.018 Received 9 May 2018; Received in revised form 13 July 2018; Accepted 14 July 2018 ⁎ Corresponding authors. E-mail addresses: augonzalez@gdl.cinvestav.mx (A. González-Casillas), laparra@gdl.cinvestav.mx (L. Parra), ldmartin@gdl.cinvestav.mx (L. Martin), framos@gdl.cinvestav.mx (F. Ramos). Biologically Inspired Cognitive Architectures xxx (xxxx) xxx–xxx 2212-683X/ © 2018 Elsevier B.V. All rights reserved. Please cite this article as: Gonzalez Casillas, A., Biologically Inspired Cognitive Architectures (2018), https://doi.org/10.1016/j.bica.2018.07.018