A neurobiological model of visual attention and invariant pattern recognition based on dynamic routing of information

By Barbara Olshausen, Charles H. Anderson, D. V. Van Essen. 1993

We present a biologically plausible model of an attentional mechanism for forming position- and scale-invariant representations of objects in the visual world. The model relies on a set of control neurons to dynamically modify the synaptic strengths of intracortical connections so that information from a windowed region of primary visual cortex (V1) is selectively routed to higher cortical areas. Local spatial relationships (i.e., topography) within the attentional window are preserved as information is routed through the cortex. This enables attended objects to be represented in higher cortical areas within an object-centered reference frame that is position and scale invariant. We hypothesize that the pulvinar may provide the control signals for routing information through the cortex. The dynamics of the control neurons are governed by simple differential equations that could be realized by neurobiologically plausible circuits. In preattentive mode, the control neurons receive their input from a low-level “saliency map” representing potentially interesting regions of a scene. During the pattern recognition phase, control neurons are driven by the interaction between top-down (memory) and bottom-up (retinal input) sources. The model respects key neurophysiological, neuroanatomical, and psychophysical data relating to attention, and it makes a variety of experimentally testable predictions.

Subfields: Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

Citations in the corpus

Perceiver: General Perception with Iterative Attention (2021, background): “There is widespread evidence that it plays an important role in human vision (e.g. Felleman & Essen 1991; Olshausen et al. 1993; Lollo et al. 2000), which is characterized by limited bandwidth input streams (Wolfe et al., 2006).”
Squeeze-and-Excitation Networks (2017, background): “Attention can be viewed, broadly, as a tool to bias the allocation of available processing resources towards the most informative components of an input signal [17, 18, 22, 29, 32].”
Dynamic Routing Between Capsules (2017, background): “Olshausen et al. [1993] improves upon Hinton [1981b] dynamic connections and presents a biologically plausible, position and scale invariant model of object representations.”
On the Binding Problem in Artificial Neural Networks (2020, methodology): “Several mechanisms have been proposed that range from static binding using conjunction cells (Ghose and Maunsell, 1999) to dynamic information routing through dedicated circuitry (Olshausen et al., 1993; Zylberberg et al., 2010) or attention using common location tags (Reynolds and Desimone, 1999; R”
Selective Kernel Networks (2019, background): “It biases the allocation of the most informative feature expressions [16, 17, 24, 28, 31] and simultaneously suppresses the less useful ones.”
Learning Generative Models with Visual Attention (2013, methodology): “[13], whereas the right panel shows a graphical diagram of our proposed generative model with an attentional mechanism.”
Toward an Integration of Deep Learning and Neuroscience (2016, background): “Other proposed mechanisms include speciﬁc circuits for attention-dependent signal routing ( Anderson and Van Essen, 1987; Olshausen et al., 1993).”
The Packet Switching Brain (2011, unknown): “In Olshausen, Anderson, and Van Essen ʼ s (1993) model, dynamic routing circuits normalize the view of an object to a canonical size and location, obviating the need for a complete tiling of retinal space with identical sets of feature detectors.”
Computational visual attention systems and their cognitive foundations: A survey (2010, methodology): “Examples are the dynamic routing circuit [Olshausen et al. 1993], and the models MORSEL [Mozer 1987], SLAM (SeLective Attention Model) [Phaf et al.”
A Neurodynamical Model of Visual Attention: Feedback Enhancement of Spatial Resolution in a Hierarchical System (2001, background): “The gating mechanisms of Olshausen et al. (1992) or Salinas and Abbott (1997) could be integrated into our system by replacing the recognition module with these more realistic neural implementations to obtain translation, rotation, and scaling invariance.”
Squeeze-and-Excitation Networks (2017, background): “preted as a means of biasing the allocation of available computational resources towards the most informative components of a signal [46], [47], [48], [49], [50], [51].”