The Hubel–Wiesel (HW) model is a classical neurobiological model for explaining the orientation selectivity of cortical cells. However, the HW model still has not been fully proved physiologically, and there are few concise but efficient systems to quantify and simulate the HW model and can be used for object orientation detection applications. To realize a straightforward and efficient quantitive method and validate the HW model’s reasonability and practicality, we use McCulloch-Pitts (MP) neuron model to simulate simple cells and complex cells and implement an artificial visual system (AVS) for two-dimensional object orientation detection. First, we realize four types of simple cells that are only responsible for detecting a specific orientation angle locally. Complex cells are realized with the sum function. Every local orientation information of an object is collected by simple cells and subsequently converged to the corresponding same type complex cells for computing global activation degree. Finally, the global orientation is obtained according to the activation degree of each type of complex cell. Based on this scheme, an AVS for global orientation detection is constructed. We conducted computer simulations to prove the feasibility and effectiveness of our scheme and the AVS. Computer simulations show that the mechanism-based AVS can make accurate orientation discrimination and shows striking biological similarities with the natural visual system, which indirectly proves the rationality of the Hubel–Wiesel model. Furthermore, compared with traditional CNN, we find that our AVS beats CNN on orientation detection tasks in identification accuracy, noise resistance, computation and learning cost, hardware implementation, and reasonability.

How specific features of the environment are represented in the mammalian brain is an important unexplained mystery in neuroscience. Visual information is considered to be captured most preferentially by the brain. As one of the visual information elements, motion direction in the receptive field is thought to be collected already at the retinal direction-selective ganglion cell (DSGC) layer. However, knowledge of direction-selective (DS) mechanisms in the retina has remained only at a cellular level, and there is a lack of complete direction-sensitivity understanding in the visual system. Previous studies of DS models have been limited to the stage of one-dimensional black-and-white (binary) images or still lack biological rationality. In this paper, we innovatively propose a two-dimensional, eight-directional motion direction detection mechanism for grayscale images called the artificial visual system (AVS). The structure and neuronal functions of this mechanism are highly faithful to neuroscientific perceptions of the mammalian retinal DS pathway, and thus highly biologically reasonable. In particular, by introducing the horizontal contact pathway provided by horizontal cells (HCs) in the retinal inner nuclear layer and forming a functional collaboration with bipolar cells (BCs), the limitation that previous DS models can only recognize object motion directions in binary images is overcome; the proposed model can solve the recognizing problem of object motion directions in grayscale images. Through computer simulation experiments, we verified that AVS is effective and has high detection accuracy, and it is not affected by the shape, size, and location of objects in the receptive field. Its excellent noise immunity was also verified by adding multiple types of noise to the experimental data set. Compared to a classical convolutional neural network (CNN), it was verified that AVS is completely significantly better in terms of effectiveness and noise immunity, and has various advantages such as high interpretability, no need for learning, and easy hardware implementation. In addition, activation characteristics of neurons in AVS are highly consistent with those real in the retinal DS pathway, with strong neurofunctional similarity and brain-like superiority. Moreover, AVS will also provide a novel perspective and approach to understanding and analyzing mechanisms as well as principles of mammalian retinal direction-sensitivity in face of a cognitive bottleneck on the DS pathway that has persisted for nearly 60 years.

The perception of motion direction is essential for the survival of visual animals. Despite various theoretical and biophysical investigations that have been conducted to elucidate directional selectivity at the neural level, the systemic mechanism of motion direction detection remains elusive. Here, we develop an artificial visual system (AVS) based on the core computation of the Hassenstein–Reichardt correlator (HRC) model for global motion direction detection. With reference to the biological investigations of Drosophila, we first describe a local motion-sensitive, directionally detective neuron that only responds to ON motion signals with high pattern contrast in a particular direction. Then, we use the full-neurons scheme motion direction detection mechanism to detect the global motion direction based on our previous research. The mechanism enables our AVS to detect multiple directions in a two-dimensional view, and the global motion direction is inferred from the outputs of all local motion-sensitive directionally detective neurons. To verify the reliability of our AVS, we conduct a series of experiments and compare its performance with the time-considered convolution neural network (CNN) and the EfficientNetB0 under the same conditions. The experimental results demonstrated that our system is reliable in detecting the direction of motion, and among the three models, our AVS has better motion direction detection capabilities.

The convolutional neural network (CNN) was initially inspired by the physiological visual system, and its structure has become increasingly complex after decades of development. Although CNN architectures now have diverged from biological structures, we believe that the mechanism of feature extraction in the visual system can still provide valuable insights for enhancing CNN robustness and stability. In this study, we investigate the mechanism of neuron orientation selectivity and develop an artificial visual system (AVS) referring to the structure of the primary visual system. Through learning on an artificial object orientation dataset, AVS acquires orientation extraction capabilities. Subsequently, we employ the pre-trained AVS as an information pre-processing block at the front of CNNs to regulate their preference for different image features during training. We conducted a comprehensive evaluation of the AVS–CNN framework across different image tasks. Extensive results demonstrated that the CNNs enhanced by AVS exhibit significant model stability enhancement and error rate decrease on noise data. We propose that incorporating biological structures into CNN design still holds great potential for improving overall performance.

Artificial intelligence, with its remarkable adaptability, has gradually integrated into daily life. The emergence of the self-attention mechanism has propelled the Transformer architecture into diverse fields, including a role as an efficient and precise diagnostic and predictive tool in medicine. To enhance accuracy, we propose the Double-Attention (DA) method, which improves the neural network’s biomimetic performance of human attention. By incorporating matrices generated from shifted images into the self-attention mechanism, the network gains the ability to preemptively acquire information from surrounding regions. Experimental results demonstrate the superior performance of our approaches across various benchmark datasets, validating their effectiveness. Furthermore, the method was applied to patient kidney datasets collected from hospitals for diabetes diagnosis, where they achieved high accuracy with significantly reduced computational demands. This advancement showcases the potential of our methods in the field of biomimetics, aligning well with the goals of developing innovative bioinspired diagnostic tools.