Members of DPAG are part of an international collaborative team that has published new research in Proceedings of the National Academy of Sciences.
Modern, high sensitivity, magnetic resonance (MR) imaging now allows non-invasive methods to resolve how the human brain responds to the three-dimensional structure of the visual world. A new paper in PNAS from an international collaborative team shows how these complex MR-signals can be resolved into basic elements of processing that are activated when human volunteers look at visual patterns that systematically change in perceived depth.
Hard problems are often solved by breaking them down into a set of smaller problems, each of which is individually easier to solve. This approach, sometimes called "divide and conquer", underlies the efficiency of many processes in modern business and manufacturing and is responsible for the surprising performance of modern artificial intelligence systems. Our own brains also use this strategy. Unsurprisingly, if a path to a solution is available, biological evolution and lifetime learning will discover it. Modern neuroscientific studies have contributed substantially to the identification of basic units that support the 'divide and conquer' approach.
For biological vision, the basic units of processing are called receptive fields, because they were initially defined by a limited surface area or field over which the neuron is receptive. Stimuli in the receptive field cause the neuron to respond with electrical impulses called action potentials, whereas stimuli outside this area cause no response. In the initial stages of visual processing in the cerebral cortex, the receptive fields are responsive not only to where the visual stimulus is positioned, but also its shape, orientation, movement, colour, texture and other properties.
Recently the quality of non-invasive, magnetic resonance (MR) signals available from human cortex has become good enough to allow the identification of units of processing that resemble receptive fields. The new work extends the mapping of these MR-based receptive fields into the third dimension of space, which is the distance away from the viewpoint of the human subject. Specifically, the team tested binocular depth, which you may have experienced at the 3-D movies using special 3-D glasses that send different images to the left and right eyes. Using a specialised 3-D viewer, MR-imaging was used to measure the brain responses of volunteers, as the depth of visual patterns was varied systematically. Computational data analysis then resolved the complex MR responses into a set of basic receptive fields for binocular depth. These physiological receptive fields form units of processing used by the human visual system for the computation of binocular depth.
New or changed methods always need validation. The new MR-based measurements were made in humans using a blood signal that is only indirectly related to neuronal activity. The new measures were compared with earlier measurements of brain activity made in macaque monkeys, which directly sampled the electrical activity of single neurons in the visual cortex. Since macaques have binocular vision very similar to ours,
the identification of many similarities in the two data sets is encouraging but there were also differences that need further investigation. In particular, the human MR-based responses show a greater sensitivity to the relative depth difference between visible features.
Large-scale brain imaging studies have established benchmarks for the anatomical structure of typical, healthy human brains. The focus is now shifting to finding benchmarks for physiological measures of brain function, specifically here the visual receptive fields for binocular depth. These new benchmark measures will support the assessment of changes in visual function in the large number of clinical cases in which vision becomes lost or disrupted, due to problems in the central parts of the brain rather than problems within the eyes themselves.
Professor Andrew Parker comments, ‘It was a huge pleasure to work closely with my co-authors to bring meaningful structure to this large and complex dataset’.
The research team are:
Andrew J. Parkera,b,c,1, Ivan Alvarezd,e ID, Alessandro Mancarie,f ID, I. Betina Ipd,e ID, Kristine Kruga,b,c ID, and Holly Bridged,e ID
a Department of Sensory Physiology, Institute of Biology, Otto von Guericke University, 39120 Magdeburg, Germany
b Department Physiology, Anatomy and Genetics, University of Oxford, Oxford, OX1 3PT, United Kingdom
c Leibniz-Institute for Neurobiology, 39120 Magdeburg, Germany
d Oxford Centre for Functional Magnetic Resonance Imaging of the Brain, Centre for Integrative Neuroimaging, University of Oxford, Oxford, OX3 9DU, United Kingdom
e Nuffield Department of Clinical Neurosciences, University of Oxford, Oxford, OX3 9DU, United Kingdom
f Department of Translational Research on New Technologies in Medicine and Surgery, University of Pisa, 56124 Pisa, Italy
The full article, ‘Receptive fields from single-neuron recording and MRI reveal similar information coding for binocular depth’ can be accessed here