Research

Visual physiology of primate retinal ganglion cell types: Adaptation to temporal variation in light intensity, or contrast, is a fundamental property of the visual system and has been observed previously in mammalian retinal ganglion cells, including primate. In the primate lateral geniculate nucleus (LGN), magnocellular (LGNm), but not parvocellular (LGNp) relay cells showed contrast adaptation. This result seems counterintuitive since the parvocellular pathway mediates achromatic spatial resolution and is critical for form perception.  In this ongoing study, we measure contrast adaptation in identified LGNp-projecting midget in LGNm-projecting parasol retinal ganglion cells in a light adapted in vitro preparation of the macaque monkey retina for the first time.

  • Kim, Y.J., Packer, O.S., Detwiler, P.B., & Dacey, D.M. (2019). Achromatic contrast adaptation parasol and midget ganglion cells of the macaque monkey retina. Investigative Ophthalmology & Visual Science, 60(9), 5276. PDF

Neural circuitry of direction selectivity in the primate retina: Starburst amacrine cells are an intensively studied inhibitory retinal interneuron that has been implicated as a key component in neural mechanism for direction selectivity in retinal ganglion cells. A number of lines of evidence from previous work in mouse and rabbit retina suggest that starburst dendrites can function independently as local motion-detectors and make highly selective inhibitory synaptic output to direction selective ganglion cells. In this line of investigation we have initiated a study of the starburst amacrine cells in the primate retina where nothing is known about direction selectivity. 
       How do we know that starburst dendrites can function independently as local motion detectors? This was shown in an breakthrough study that used 2-photon calcium imaging in the dendrites of rabbit starburst cells (Euler, Denk & Detwiler, Nature 2002, PDF). This method allows us to study the light response in selected dendrites instead of just recording the light response at the cell body. In this study, rabbit starburst cells were filled with a fluorescent calcium indicator like OGB (Oregon Green Bapta). When the cell was excited by moving visual stimuli (e.g,  bars, or radial gratings), calcium channels open and calcium goes up in the cell, OGB binds more calcium and fluorescence increases. This surprising study showed that individual dendrites would respond to different motion directions depending on their location around the starburst tree.

(Note: Distinctive dendritic morphology of starburst amacrine cells in the macaque monkey retina. Four examples of ON-center starburst cells filled with Alex FLOUR 568 and OGB.)
       We know that starburst amacrine cells existed in the primate (see above), however as i mentioned earlier they had never been targeted for functional studies. In current ongoing study, we look at direction selectivity in the dendrites we have also started to use 2-photon calcium imaging to test for direction selectivity of starburst dendritic tips. We found that starburst amacrine cells in the primate macaque monkey show strong direction selectivity.

Connectomic analysis of human/non-human primate foveal cells and circuits: “Connectomics” is a broad term to refer to recently developed methods of serial section electron microscopic reconstruction of neural cells and circuits. First, recent advances in microscopy now permit thousands of images to be collected from a tissue “block-face” during computer controlled sectioning from inside of a scanning electron microscope (SBEM). Second, advances in computer power and software permit large Terabyte volumes to be anayzed with sophisticated tools for 3D reconstruction. Finally, deep learning and convolutional neural network approaches are being developed that permit reconstructions to be semi-automated. Taken together these methods have created a new field usually referred to as connectomics.   

(Left: Reconstructed Cone Array in the Human Fovea. In this image you see a reconstruction of the synaptic endings of the cones in the very center of the human fovea (red cone is at the foveal center) the long processes are the axons of the cones also known as Henle fibers. Each cone then terminates in a bulb-like swelling that contains around 25 synaptic ribbons that permit each cone to make synaptic output to about a dozen bipolar cell types and to many horizontal cell interneurons as well. In addition you can see that each cone projects small laterally extending processes. These are the cone telodendria. At the tip of each telodendria is a gap junction (to permite direct electrical communication) with a neighboring telodendria from an adjacent cone pedicle. In the human foveal center all the cones are either L or M (Long or Middle wavelength sensitive). S cones are also present just beyond the array of cones shown here.)

The great advantage of the connectomics approach is that we can directly quantify the detailed excitatory and inhibitory synaptic connections between cells that define neural circuits. These connections allow us to interpret physiological data and to model circuits with biologically realistic neural simulations. I am currently using the connectomics approach to address several questions about the organization of the human fovea and macaque monkey circuitry underlying direction selectivity. 

Back to Top ↑

 

 

Vision and cognitive neuroscience: The goal of this research was to advance understanding how real-world (natural scene) visual information is encoded within the human brain and perceived by the human observer. The natural scene is composed both color and luminance contrast at different sizes and orientations. Much of our understanding of visual processing, however, comes from studies focusing on the aspect of color alone or luminance alone. I am interested in studying the interactions between color and luminance contrast that occurs in everyday viewing of the world.  To study these, I employ the behavioral testing of human vision (psychophysics), brain stimulation (TMS), and computational modeling.

Processing of colorful natural scenes: Natural scenes contain both color and luminance contrast at different sizes and orientations that are sometimes spatially overlaid and sometimes not, as in shadows and shading (see example below). The study of the interactions between color and luminance contrast is essential to the understanding of realistic visual processing that occurs in our normal perception of the world. Much of our understanding of these interactions is well established at detection threshold, but relatively little is known at perceived contrast levels.

Note: Natural scenes (original image) contain both luminance (luminance-only) and color (roughly isoluminant image) (Note that the original image is from McGill colour image database)

At perceived contrast levels, interesting findings are revealed in which the relative alignments of border positions between color and luminance (black and white) contrast play an important role that potentially reveals two different mechanisms: color suppression when the luminance and color borders coincide to demarcate an object boundary, and color enhancement when they do not coincide but are shading and shadowing effects. These results suggest a novel way of understanding color-luminance interaction in natural scenes. These border effects are important in consideration of how color and luminance contrast co-occur in natural scenes: In natural scenes, aligned color and luminance borders generally com from changes in surface reflectance associated at object boundaries, however luminance borders with no associated color change typically come from shadows and shadings.

  • Kim, Y.J. & Mullen, K.T. (2016). Effect of overlaid luminance contrast on perceived color contrast: shadows enhance, borders suppress. Journal of Vision, 16(11):15, 1-14. PDF

Employing Transcranial Magnetic Stimulation (TMS), the role of hMT+ in color and luminance motion perception was explored, finding that while are hMT+ is part of a pathway common to processing the motion of (isoluminant red-green) color and luminance patterns, it is not involved in their detection.

  • Kaderali, S., Kim, Y.J., Reynaud, A., & Mullen, K.T. (2015). The role of human brain area hMT+ in the perception of global motion investigated with repetitive transcranial magnetic stimulation (rTMS). Brain Stimulation, 8, 200-207. PDF

Contrast normalization mechanism in luminance vision: When a pattern of broad spatial content is viewed by a human observer, the multiple spatial components in the pattern stimulate detecting-mechanisms that suppress each other. This mutual suppression in neural activity is referred to as contrast gain control and has been well characterized by several models of contrast normalization, in which the activity of a neural detector at an early visual stage (V1) is divided by the pooled activities of a number of neural detectors forming a contrast gain control pool, in a so-called “divisive normalization” process. Through this modification process, neurons in the visual cortex can efficiently respond to a wide range of contrasts and maintain stimulus selectivity. The contrast gain control process therefore is one of the most fundamental efficient coding mechanisms of the visual system and the underlying neural mechanisms have been intensively studied by nonlinear systems analysis using simple stimuli. However, relatively little was known about its response tuning properties with naturalistic (broadband) stimuli. The broadband stimuli drive a range of spatiotemporally-tuned neurons (also thought of as filters) and activate the suppressive interactions between them to form contrast gain control mechanisms in early visual processing. I determined how the action of these mechanisms depends on stimulus bandwidth as well as other spatiotemporal properties. I discovered that suppressive interactions stem from locally-tuned and anisotropically-weighted gain-control pools in multiple spatio-temporal dimensions and demonstrated a new view of “sustained” and “transient” mechanisms and has important implications for understanding the circuitry of visual cortex.

  • Kim, Y.J., Haun, A.M., & Essock, E.A. (2010). The horizontal effect in Suppression: Anisotropic overlay and surround suppression at high and low speeds. Vision Research, 50, 838-849. PDF
  • Essock, E.A., Haun, A.M., & Kim, Y.J. (2009). Anisotropy of orientation-tuned suppression that matches the anisotropy of typical natural scenes. Journal of Vision, 9(10):35, 1-15. PDF

 

Contrast normalization mechanism in color vision: I have extended my research on the contrast gain control mechanism in luminance vision to include color vision. My modeling approach has continued to focus on the role of contrast gain control mechanisms, trying to parse the differential role of monocular and binocular sites of contrast gain control mechanisms in color and luminance vision. These two types of contrast gain control processes have been well characterized by models of cross-orientation masking in luminance contrast, color contrast, and color and luminance contrast in combination, and in terms of their temporal dynamics.

  • Kim, Y.J. & Mullen, K.T. (2015). The dynamics of cross-orientation masking at monocular and dichoptic sites. Vision Research, 116, 80-91. PDF
  • Mullen, K.T., Kim, Y.J., & Gheiratmand, M. (2014). Contrast normalization in color vision: the effect of luminance contrast on colour contrast detection. Scientific Report, 4, 4285, 1-7. PDF
  • Kim, Y.J., Gheiratmand, M., & Mullen, K.T. (2013). Cross-orientation masking in human color vision: Application of a two-stage model to assess dichoptic and monocular sources of suppression. Journal of Vision, 13(6):15, 1-14. PDF

Contrast sensitivity in color and luminance vision: My previous research focused on characterizing the monocular and binocular cone-contrast sensitivity functions using the quick Contrast Sensitivity Function approach (qCSF) for red-green, blue-yellow, and achromatic luminance stimuli to get normative data set.

  • Kim, Y.J., Reynaud, A., Hess, R.F., & Mullen, K.T. (2017). A normative dataset for the clinical assessment of achromatic and chromatic contrast sensitivity using a qCSF approach. Investigate Opthalmology & Visual Science, 58(9), 3628-3636. PDF

Neural plasticity: Binocular interactions using a dichoptic movie-viewing paradigm to manipulate the interocular balance has been studied, and found that an interocular imbalance in achromatic contrast and not chromatic contrast drives plastic changes in boulcar interaction.

  • Zhou, J., Reynaud, A., Kim, Y.J., Mullen, K.T. & Hess, R.F. (2017). Chromatic and achromatic monocular deprivation produce separable changes in human ocular dominance. Proceedings of the Royal Society B: Biological Sciences, 284: 20171669 pp.1-10.1669. PDF

Back to Top ↑