No blind alleys for blindsight: multiple active pathways into extrastriate cortex



Extra-striate visual area V5/MT has a central role in processing visual information about motion and binocular depth. Neuronal signals in V5/MT have been linked to the perception of direction of motion, 3D and structure from motion (Krug, 2004). V5/MT receives inputs from a number of cortical and subcortical structures. The main source of visual input is thought to come from the retina via the lateral geniculate nucleus (LGN), through primary visual cortex (V1) directly to MT or via secondary visual cortex (V2) (Ninomiya et al. 2011). However, there is also strong evidence for a direct pathway from the LGN to V5/MT (Sincich et al. 2004) (Fig. 1).

Figure 1.

A schematic diagram of some of the key feedforward pathways into primate V5/MT 
K, koniocellular; M, magnocellular; P, parvocellular.

This pathway has been postulated to be of central importance for the remaining visual capacity in blindsight patients (Bridge et al. 2008). Patients with blindsight have lesions in V1, but can still discriminate high contrast moving stimuli seemingly without full awareness of the visual stimulus. The neural substrate for blindsight has also been investigated in monkeys with V1 lesions. A recent study blocking activity in the LGN underlined the importance of a direct pathway from the LGN to V5/MT to account for monkeys’ ability to discriminate visual stimuli within a V1 scotoma (Schmid et al. 2010).

While such studies can provide evidence for the existence of the direct pathway, the question remains what functional significance this pathway has in healthy subjects. Recently, a human fMRI study suggested that voxels in human MT+ are more strongly correlated with voxels in the LGN than with voxels in V1 (Gaglianese et al. 2012). These data suggest that the direct LGN pathway could be an important driving force for neuronal responses in MT+. However, the result could also be influenced by the similar visual magnification factors in MT and LGN, while in V1 a similar extent of the visual field occupies generally a larger area of cortex, thereby diluting any correlation between voxels of the same size in the LGN and V1.

A promising approach to understanding the contribution of the different pathways to visual processing is the functional dissection of the direct and indirect pathway as undertaken in a paper in this issue of The Journal of Physiology by Jayakumar et al. (2013). The LGN is subdivided in magno-, parvo- and koniocellular layers (Fig. 1); the neurons in each of these subcompartments show distinct response properties. Neurons in the magnocellular layers are selective for low spatial and high temporal frequencies and can respond to low contrast luminance stimuli. Neurons in the parvocellular layers preferentially respond to high spatial and low temporal frequencies and many are wavelength selective showing red–green opponency. Many neurons in the koniocellular layers are also wavelength selective showing blue-yellow opponency. Jayakumar and colleagues used S-cone isolating stimuli to selectively activate this latter pathway. They compared neuronal responses to these stimuli in visual area V5/MT before and after reversibly switching off the corresponding visual field representation in V1 through cooling.

Even when visual stimuli were entirely presented in the centre of the visual field whose representation was inactivated in V1, neuronal responses in a number of neurons in V5/MT were not affected. This suggests that such neurons received their inputs through an alternative pathway that bypasses V1, presumably directly from the LGN. Conversely, the response of other V5/MT cells was affected. Taken together, this indicates that S-cone input reaches V5/MT through both pathways. Similar results were shown for luminance modulated stimuli.

Interestingly, the timing of the responses in V5/MT for the indirect and direct pathways showed a large overlap. Conventionally, magnocellular inputs have been thought to dominate V5/MT responses serving the fast processing of visual motion information for quick action. A direct input from the LGN was thought to benefit such processing. The recent data on response times indicates that for each type of input, responses arrive through both pathways on overlapping time scales. The only partial abolition of responses in some V5/MT neurons by V1 cooling might be taken to indicate that both pathways feed into the same circuits. This underlines the functional importance of the direct pathway for preserving some visual function in blindsight.

To probe deeper into the functional significance of the different pathways to V5/MT, experiments are required that selectively and reversibly alter activity in these distinct processing streams in the behaving animal.