Visual Pathway and Cortex

Visual Cortex

The visual pathway and cortex are complex systems, and the content covers classical textbook information about cortical receptive field properties, visual cortex columns (including orientation columns), ocular dominance disparity, and how this relates to 3-D television technology.

Visual Processing in the LGN

Receptive Fields

In the LGN (Lateral Geniculate Nucleus), receptive fields are more complex than simple excitatory center and inhibitory surround models. "On" represents an increase in illumination, while "off" represents a decrease. The center is excitatory for "on" and inhibitory for "off", with the surrounds having the opposite effect.

  • On-Off Antagonistic Receptive Field: The receptive field has both "on" and "off" centers that are mutually antagonistic.

  • On: Increased brightness excites the cell.

  • Off: Decreased illumination excites the cell.

The LGN cells exhibit center-surround antagonism, responding to both increased and decreased illumination in their respective zones. Experiments using stimuli map these receptive fields, with red indicating "on" responses and blue indicating "off" responses.

X and Y Cells

Recordings made in anesthetized cats have revealed X and Y cells, which are analogous to primate parvo- and magnocellular cells, respectively, and exhibit center-surround antagonistic receptive fields.

LGN as a Faithful Relay

  • Experiments reconstruct LGN cells and map their retinal inputs.

  • Retinal ganglion cell axons form powerful synapses on the proximal dendrites of LGN cells.

  • Each retinal ganglion cell makes many connections with a single LGN cell.

The LGN is traditionally viewed as a faithful relay of retinal ganglion cell responses to the cortex. Cross-correlation analysis has been used to confirm connections between retinal and LGN cells, showing similar receptive fields.

Non-Retinal Inputs and Inhibitory Mechanisms

Only about 7% of synapses on the LGN come from the retina. Brainstem inputs modulate attention through cholinergic and neuromodulatory signals. The cortex provides 30% of the synapses to the LGN, posing a question about the necessity of these inputs for a faithful relay.

  • Inhibitory Inputs: Strong inhibition in the LGN enhances center-surround antagonism.

  • GABAergic Inhibition: Stimulation that shuts up a retinal cell also produces strong GABAergic inhibition in the LGN.

Inhibitory inputs come from cells with off-center receptive fields and are crucial for signaling contrast to the cortex.

Cortical Inputs

The primary visual cortex (V1) feeds back to the LGN, forming a loop whose function depends on cortical activity. Understanding cortical receptive fields sheds light on this feedback mechanism.

Receptive Fields in Primary Visual Cortex (V1)

Differences Between Cats and Primates

A key difference exists between cats and primates: primate LGN layers 1 and 2 project to layer 4 alpha, maintaining center-surround antagonism. In cats, cortical-type fuzzy fields are observed in the LGN.

Classical Recordings

Classical recordings from cortical cells demonstrate orientation selectivity and direction selectivity. Peri-stimulus time histograms (PSTHs) are used to map responses to bars of light moving at different orientations. Polar diagrams quantify these responses, showing strong tuning to specific orientations and directions.

Cortical cells are fussy about stimulus parameters such as bar length and velocity. Hubel and Wiesel received a Nobel Prize for their work on V1 cells, showing they prefer elongated stimuli and have orientation/direction tuning.

Simple, Complex, and Hypercomplex Cells

  • Simple Cells: Have separate on and off zones arranged side by side.

  • Complex Cells: Have overlapping on and off zones.

  • Hypercomplex Cells: Respond to short bars and have length-tuned or end-stopped responses.

Hubel and Wiesel proposed a hierarchical model: simple cells in layer 4 receive input from LGN cells, complex cells receive input from simple cells, and hypercomplex cells are found in layers 2 and 3. This model suggests that simple cells' orientation selectivity results from the convergence of LGN cells with aligned receptive fields.

Challenges to the Hierarchical Model

The hierarchical model was developed during the early days of computing and has not stood the test of time. The Alonzo Lab used cross-correlation techniques to show that cortical cells with off-center receptive fields are linked to LGN cells. However, data does not fully support the idea that elongated central zones in simple cells explain orientation selectivity.

Gaps and Inconsistencies

  • Cells are not neatly separated into distinct layers as the model suggests.

  • End-stopped cells exist across layers, not just in layers 2 and 3.

  • Inhibition is crucial for orientation selectivity.

Experiments by the Sillitoe lab show that blocking inhibition eliminates both orientation and direction selectivity, even at low stimulus contrasts. A quarter of cerebral cortex cells are inhibitory, emphasizing their functional importance.

Up-to-Date View of Visual Cortex

Sensitivity to Differences

The visual system is highly sensitive to differences and deviations from uniformity. Cortical cells respond best to stimuli that contain differences and are set up to look for texture differences. So a reexamination of cortical columns and sensitivity to differences are in order.

Experiment on Orientation Selectivity

Experiments show that cortical cells respond most strongly when the center and surround have orthogonal orientations. These cells are designed to detect differences between adjacent patches, not absolute orientations.

LGN and End-Stopping

LGN cells, though not orientation-selective, are sensitive to bar length. Layer 6 cortical feedback imposes strong end-stopping onto LGN cells. This feedback is crucial for developing sensitivity to discontinuities in the LGN, making it part of the cortical circuit.

LGN and Discontinuities

LGN cells respond better to stimuli with discontinuities in orientation or temporal dynamics, and this preference is imposed by the cortex.

Cortical feedback imposes sensitivity to discontinuities in the LGN, leading to a global function of detecting differences in the visual field.

Columnar Organization

Ocular Dominance Columns

LGN axons from different eyes terminate in separate blocks in layer 4. Cells outside layer 4 respond to one eye or the other, or to both. Horizontal sections of the cortex reveal a fingerprint-like pattern of ocular dominance columns.

Orientation Columns

Orientation columns are traditionally depicted as straight lines orthogonal to ocular dominance columns. However, in reality, orientation preferences change smoothly across the cortex, with neighboring columns having similar preferences. Within a hypercolumn, the full range of ocular dominance and orientation preferences can be found.

Pinwheel Centers

Looking down at the orientation column system, there are pinwheel centers where orientation columns converge. These centers contain non-oriented cells and have higher metabolic rates. They align with the middle of ocular dominance columns.

Stereopsis and 3-D Vision

Disparity

The two eyes each capture slightly different images of objects not at the point of fixation. This difference in location is called disparity. Some cortical cells prefer mismatched images, responding best to the disparity produced by near or far objects.

Neural mechanisms underlying stereopsis enables the perception of depth.