Neurons in the superior colliculus wake up to see things differently

Almost 60 years ago, Hubel and Wiesel described the receptive field properties of neurons in the visual cortex (Hubel & Wiesel, 1959), transforming our understanding of how the brain processes sensory input and spurring investigations that have elucidated the function, organization, development and plasticity of visual circuits. Initially focused on more visually driven species, the genetic manipulability of the mouse has allowed investigators to probe circuits in increasingly sophisticated ways. A central visual processing centre in the mouse brain is the superior colliculus (SC), a midbrain structure that processes visual, somatosensory and auditory stimuli and mediates head and eye movements. Since it receives input from the vast majority of retinal ganglion cells (RGCs), as well as top‐down inputs from the visual cortex, the mouse SC has emerged as an attractive model to investigate visual processing. However, the vast majority of these studies have been performed in anaesthetized mice, limiting insight into the native tuning properties of SC neurons. Further, a classification scheme for visual neurons in the SC lags far behind that for RGCs or neurons in the visual cortex.

In this issue of The Journal of Physiology, De Franceschi and Solomon (2018) report findings that address both of these issues, advancing our understanding of visual function in the SC. They recorded responses from SC neurons in the awake and anaesthetized states while presenting visual stimuli to compare tuning properties between each. Consistent with previous reports (Wang et al. 2010; Gale & Murphy, 2014; Ito et al. 2017), they found that in both conditions most neurons in the SC respond to both white and black spots on a grey background, and that their On and Off subfields were overlapping. Interestingly, they found differences in the response of SC neurons to drifting gratings. While no change in the proportion or magnitude of direction selectivity or orientation selectivity was observed, SC neurons in the awake state preferred higher spatial and temporal frequencies than those in anaesthetized mice. Furthermore, the authors report increased contrast sensitivity in the awake state, a visual tuning parameter previously unexplored in the SC. Intriguingly, in the awake state they found three distinct contrast tuning functions – linear, where firing rate increases with contrast; saturating, where the response plateaus at intermediate contrast; and super‐saturating, where rate peaks at intermediate and declines at higher contrasts – whereas in anaesthetized animals contrast tuning was predominantly linear. Such distinct contrast tuning functions could be achieved via neuron‐specific gain control; however, the circuit mechanisms underlying gain control in the SC remain unclear, as are the effect of anaesthesia on any such mechanism.

Next, the authors asked if eye movements made in the head‐restrained configuration might impact the determination of receptive field properties, particularly in regards to response linearity and spatial frequency tuning. To do so, they utilized infrared video oculography to estimate eye position while presenting visual stimuli. Indeed, they observed in some cases that azimuth displacements in eye position resulted in temporal shifts in spiking activity in response to a vertical drifting grating. Such shifts would alter the response rates at the first harmonic of the stimulus, impacting the calculation of a cell's spatial summation properties. To account for this, the authors cleverly performed Fourier analyses on a trial‐by‐trial basis, rather than on cumulative data, and found that this increased the numbers of neurons that exhibit linear spatial summation of drifting gratings. Intriguingly, the authors also observed a high proportion of units that exhibit linear spatial summation to contrast‐modulated gratings, especially in the anaesthetized state. While previous studies have identified linear neurons in the SC, the vast majority appear to be non‐linear, consistent with their overlapping On and Off subfield structure (Wang et al. 2010; Kay & Triplett, 2017). The authors speculate that the unequal response to black and white stimuli they observed, which has not been previously reported, could account for these discrepancies. Other factors, such as experimental condition, anaesthesia and analytical techniques may also contribute, underscoring the importance of careful design, implementation and reporting of experimental methods for inter‐study comparisons.

Finally, the authors utilized a fuzzy k‐means clustering method to classify neurons identified in the awake state based on 10 dimensions of functional response variation. Units were placed in five distinct subgroups – slow, orientation‐/direction‐selective, transient, fast untuned and small sustained – which both overlap and diverge from previous attempts to classify based on morphology, intrinsic electrophysiology or functional properties (Gale & Murphy, 2014). For instance, horizontal cells are poorly tuned to orientation, direction or speed, but prefer lower spatial frequencies, similar to the ‘fast untuned’ subgroup here. On the other hand, stellate and narrow field cells have small receptive fields and prefer high spatial frequencies, and both may fall into the ‘small sustained’ subgroup. By selecting five subgroups, the authors attempt to balance between ‘splitting’ and ‘lumping,’ while at the same time leveraging important distinctions in the awake and anaesthetized states to identify new subgroups. While this strategy falls short of a comprehensive classification, this work represents an important step towards a schema incorporating morphological, intrinsic, functional, and molecular properties. In combination with their novel findings regarding tuning in the awake state and novel analyses to account for eye movements, the authors have moved us towards a more complete understanding of visual circuitry in the SC.

Additional information

Competing interests

None declared.

Author contributions

Sole author.

Funding

This work was supported by NIH grant R01EY025627.