Human oculomotor and visual systems

Visual processing begins in the photoreceptor cells, specialised neurons of the retina (back part

of the eye). These cells are responsible for visual phototransduction, the conversion of light

into electrical signals. There are three known photoreceptor cells in the human eye; rods, cones

and photosensitive retinal ganglion cells (Figure 19).

The visual process begins at the retina and progresses to the optic nerves, optic chiasm and

optic tracts. The ipsilateral and decussating pathways are demonstrated in Figure 20 as well as

the retinotopic visual scene projections to the primary visual areas (striate) in the occipital

cortex. The optic chiasm is formed by the crossing of the optic nerves, with axons from the

medial (nasal) retina decussating and continuing as the contralateral optic tract. Lateral

(temporal) retina axons remain ipsilateral (do not cross at the chiasm) and feed into the

ipsilateral optic tract. During primate vision, the majority of fibres from the medial half of the

retina cross while all the fibres from the lateral half of the retina remain ipsilateral,

enabling binocular vision (Reamington, 2012). Therefore, signals from the left visual field are

passed to the right visual primary cortex ipsilateraly, via the temporal hemiretina of the right

eye, or contralaterally, via the nasal hemiretina of the left eye, and vice versa for the right visual

Figure 20. Segregation of the human visual field (reproduced from Betts et al., 2013).

There are two major pathways within the human brain which process visual information. The

largest pathway is the retina-geniculate-striate pathway which passes information from the

retinal neurons in the eye to the optic nerve and then onto the primary visual cortex in the

occipital lobe via the lateral geniculate nuclei (LGN) of the thalamus (Burman & Wurtz, 2008).

The LGN is part of thalamic structures and is a relay centre and proposed ‘early gatekeeper’ in

Additionally, the LGN receives feedback connections from the primary visual and striate

cortices (Briggs & Usrey, 2011). The second pathway is the tectopulvinar pathway. This

pathway passes information from the optic nerve onto the superior colliculus (SC), a midbrain

structure, next to the pulvinar nucleus in the thalamus, and lastly the visual cortex with primary

projections to the parietal and temporal lobes (Kolb & Whishaw, 2016) (Figure 21).

Figure 21. Diagram of the flow of visual information from the eye into the brain (reproduced from Kolb and Whishaw, 2016, p.234). The optic nerve consists of two branches. The first branch projects to the LGN in the thalamus (geniculostriate system) and the second branch projects to the SC (tectopulvinar system). The LGN pathway enters the occipital cortex and the information is passed to other visual regions in the temporal and parietal cortices. The SC pathway leads to the pulvinar and, again, onto the temporal and parietal cortices.

After initial visual processing in the primary visual cortex, information is transmitted to

secondary and association visual areas in the occipital lobes, collectively termed the striate

cortex (Standring, 2015). The majority of pathways which mediate transference between the

primary and secondary association areas are part of two main streams; the dorsal and the ventral

stream (Goodale & Milner, 1992). In the most simple form, the dorsal stream (occipitoparietal

pathway) projects visual information to the parietal cortex and has been coined the 'where is it'

processing. The ventral stream (occipitotemporal pathway) projects visual information to the

temporal cortex and has been coined the 'what is it' pathway as it is responsible for colour, form,

and identity of an object (Goodale & Milner, 1992). The occipitoparietal and occipitotemporal

pathways are output pathways from the visual cortex and receive numerous top-down input

from distributed brain regions (Kravitz, Saleem, Baker, Ungerleider & Mishkin, 2013).

Goodale and Milner (1992) proposed that the main difference between the two streams was the

function in output systems. Both the dorsal and ventral streams process information about the

structure and location of an item but the two separate streams process and disseminate the

information differently. The ventral stream generates perceptual representations from visual

inputs concerning an object's structure and location (vision for perception). The dorsal stream

mediates real-time visual actions, such as reaching and grasping, by constantly processing

visual information about an object and creating and updating the egocentric coordinates of the

object within the environment (vision for action). Goodale and Milner (1992) therefore

proposed a division between the vision for perception and vision for conducting actions.

Empirical support for the ‘two visual system hypothesis’ is robust and heavily based on

neuropsychological dissociations for object and spatial vision. For example, while some brain-

injured individuals present with visual agnosia, the inability to recognise everyday objects,

others present with optic ataxia, the inability to control visually orientated hand movements

(Goodale, Milner, Jakobson & Carey, 1991; Martinaud et al., 2012; McIntosh, Mulroue,

Blangero, Pisella & Rossetti, 2011). However, evidence supporting this apparent double

dissociation has been challenged. For instance, Pisella, Binkofski, Lasek, Toni and Rossetti

(2006) proposed that individuals in case studies were not administered an extensive visual

perception assessment which raises questions concerning the validity of Goodale and Milner’s

(1992) conclusions. Milner and Goodale (2008) rebutted these challenges and suggested that some controversies around the ‘two visual system hypothesis’ were based on an imprecise

reading of some of the more subtle details of the model. The authors clarified the terms ‘vision for perception’ and ‘vision for action’ and provided a fuller account of the processing

characteristics for these two kinds of vision. Nevertheless, there is still uncertainty surrounding

the ‘two visual system hypothesis’. For example, Michel and Henaff (2004) argued against the

dichotomy of the model, contending that the functional double dissociation, demonstrated by

brain-injured individuals, could stem from aberrant central or peripheral vision rather than

impairments in the vision for perception and vision for conducting actions. Rossetti, Pisella

and McIntosh (2017) suggested that it is no longer tenable to assume independence between

visuomotor and visual perception functions based solely on the double dissociation between

optic ataxia and visual agnosia in brain-injured individuals. The authors highlighted the

importance of taking into account the perception-action circle as a functional system.

Compelling experimental and tractography research mapping white matter connectivity has

shown a myriad of interconnected pathways, opposing the supposed isolated dual visual system

theory. More than likely, the two systems approach is an oversimplification and there is cross- talk between the pathways (Budisavljevic, Dell’Acqua & Castiello, 2018). Indeed, a large

contemporary fMRI study utilising data from the Human Connectome Project suggested that

the functional human visual system may incorporate three, not two, cortical pathways (Haak &

Beckmann, 2018), highlighting the complexity of the system and the reality that there are, more

than likely, multiple visual circuits.