The potential importance of neural recurrence for motion perception
A natural intuition emerges when one becomes versed in the sensory transduction mechanisms of the human brain: information that was once observer-independent can be continually translated and interpreted along hierarchical processing streams so as to eventually represent that sensory stimuli by way of a unified subjective experience. In a conceptually similar manner to the refinement of earth-gems through the lengthy process of sanding, buffing and polishing, spatiotemporal patterns of photon absorption by photoreceptors in the retina follow a pathway to higher cortical areas each compartmentally necessary for identifying the edges, color, gestalt grouping, and motion embedded in those patterns. As such, it appears as though information evolves as it moves through each cortical processing unit until it reaches a sufficiently coherent threshold for awareness. However, adroit neuroscience experiments have revealed a fault in this intuition; subjective awareness of higher order information, at least within the visual pathways, depends on recurrence by way of neural backprojections across cortical areas thought to be solely dedicated to specific components of visual-percept information evolution.
Area V1 in the human brain has been classically associated with first stage of cortical processing of visual information (Panizza, 1855)—containing a complete map of the visual field. Area V5, also known as MT, the suspected recipient of pre-processed visual information from V1, is thought to be responsible for the perception of motion (Zeki, 1991). Apart from the neural recordings in or observation of blood flow to these areas that speak to the correlation of their respective neural activity with presentation of visual stimuli, it is also possible to query a human’s subjective experience after artificial, spatially-selective neuromodulation to these regions by way of transcranial magnetic stimulation (TMS). In an affirmative demonstration of functional specialization within key visual regions, TMS applied to V1 enforces the percept of a stationary light-ball flash known as a phosphene (Marg, 1994) whereas application to V5/MT enforces the percept of moving phosphenes (Hotson, 1994). The generation of a static phosphene by stimulation of V1 seems compatible with this region’s role in elementary visual processing. However, the creation of a moving phosphene after stimulation of MT is a bit more puzzling—if MT is higher up on the information evolution stream of hierarchical visual processing, then how could MT elicit the percept of a phosphene, the more elementary component of the holistic percept? It would seem as though V1’s processing capabilities would need to be recruited in order to fully form this percept—a step-back in the supposed hierarchical processing stream.
A clever experiment spearheaded by Pascual-Leaon (2001) provides an insight that both resolves this “step-back” confusion and dissolves the initial natural intuition of independent, feed-forward nodes along the visual processing stream. With the working hypothesis that fast backprojections from MT to V1 are necessary for visual awareness of moving stimuli, Pascual-Leone and Walsh (2001) applied TMS to both areas with varying temporal delays so as to examine the effects of disrupting the information transfer between the two regions. Specifically, the research duo hypothesized that “if the V5 to V1 backprojection is critical for awareness, disruption of activity in V1 at the time of arrival off feedback should interfere with the perception of attributed encoded by the [V5/MT] area” (Pascual-Leone, 2001). To begin with, the researchers identified the stimulation thresholds in both V1 and V5 necessary for creating a stationary and moving phosphene, respectively. This allowed for the researchers to use two TMS coils so as to, if applied with no temporal day (i.e at the same time), create overlapping phosphenes—one stationary and one moving.
Their data revealed that sub-phosphene-threshold stimulation of V1 prevents the V1-induced phosphenes from appearing alongside the V5-induced-phosphene when the two stimulations are temporally synchronous. When TMS was applied to V1 before V5, there was never an effect on the perceived movement of the phosphene, regardless of the temporal asynchrony. This finding is in line with the natural intuition posited earlier– if V1 is responsible for creating a phosphene and MT for its movement, then stimulating V1 before MT would be compliant with the natural order of information processing. However, when V5 stimulation occurred before V1 sub-phosphene-threshold stimulation, subjects collectively reported a decrease in the quantity and quality of the phosphene (or its disappearance altogether) as compared to a pure V5 moving-phosphene-stimulation. Specifically, all subjects in this study reported that this stimulation condition (V1 stimulation 45ms after V5), in the majority of trials, created a stationary phosphene instead of a moving one. This finding would insinuate that V1 is somehow associated with the V5-induced moving phosphene.
Previous research has suggested that V5 projects to V1 over a short time course, such as the ones utilized in the Pascual-Leone(2001) experiment(Heller, 1995; Nowak, 1997). Evidently, these V5 to V1 backprojections are necessary for the visual percept of properties that were naively assumed to be local to V5. This conclusion is corroborated by the case study of a blind patient, G.Y., who had a lesion that primarily affected V1. G.Y., while receiving TMS stimulation to V5, reported no forced percept of a moving phospene, yet retained the interesting percept of movement without visual perception (gnosanopsia) (Zeki, 1998). Additionally, as revealed by fMRI, G.Y.’s MT recruited blood-flow for neuronal activity whenever he was presented with moving stimuli (Zeki, 1998). Dubbed Riddoch syndrome, this phenomenon of gnosanopsia illustrates the importance of V1 in visual awareness, even if activity is observed in V5/MT.
This evidence collectively suggests that activity in V5/MT that is permitted to propagate to V1 forms a unified visual percept of a moving visual object. In the absence of a functional V1, the percept of motion can be retained; artificially modulating V1 after MT activity will prevent the unification of motion and raw visual percept. In accordance with such findings, Bullier (2001) suggest that areas V1 and V2 may serve as “active blackboards” that integrate the results of computations performed in higher order areas, at least for the early stages of processing—essentially resulting in an area that can bind all relevant features of a stimulus into a unified percept. Further support of this theory comes by way of Lamme et al. (2000) who showed that the order in which cortical areas are activated does not necessarily correspond to that predicted by the hierarchical organization of areas. Such suggestions have a theoretical basis that may find its roots in subcortical-cortical circuitry. It has been shown that magnocellular (M) cells from the LGN reach V1 20ms earlier than parvocellular (P) cells (Nowak, 1995). This systematic delay in processing could allow for briefly orthogonal computational avenues, followed by a reintegration of preprocessed information (Bullier, What is fed back?). Specifically, if computation pertinent to M cells can be accomplished in higher order visual areas and then the resultant information, stored as a spatiotemporal patterns of firing, could be backprojected back to earlier visual areas like V1 to integrate with P cell activity. Thereby, it would seem as though neurons in V1 are recycled after their initial early-visual processing to hold a final percept that mysteriously yields subjective, conscious awareness.
Additional, somewhat contradictory, evidence suggest that area MT and V1/V2 are routinely activated at the exact same time in macaques (Raiguel, 1989). This evidence posits that it could be theoretically plausible that conscious awareness of a unified percept is not created by any one isolated region, but instead the temporally synchronous activation of each isolated region, allowing for an emergent, non-physical phenomenon to represent each individual component as part of a unified whole. If such a postulate were true, the Pascual-Leone (2001) results could be interpreted as follows: TMS stimulation to V1 after MT prevented the maintenance of V1’s contributions to a unified percept and resulted in the common experience of the disappearance of a percept altogether. Furthermore, since TMS induces a relatively widespread cortical modulation (Fuggetta, 2005), it could be the case that TMS to V1 also modulated V5 activity and vice-versa, further confounding the interpretations. Lastly, the percept of a moving phosphene after V5 stimulation like in Hotson (1994) could be explained by a propagation of stimulation to V1, allowing, eventually, the temporally synchronous activity in V5 and MT—producing a moving phosphene.
In order to resolve these two potential explanations for the dependencies between V1 and V5 for conscious motion perception (recurrent backprojections to V1 for integration vs. temporally synchronous emergent integration) and the potential for mistargeted neuromodulation, I would like to propose an experiment where the cortical targets for highly-specific modulation includes the white-matter between theV1 and V5. The first stage would be to complete a functional localizer with fMRI that selectively identifies the MT and V1 in a human. Second, a high-resolution diffusion weighted image would be collected. Using the MT and V1 regions of interest, one could use diffusion tensor techniques (Descoteaux, 2009) to probabilistically track the projections of voxels in MT that reach V1. The end result would be the identification of an isolated tract of white matter that specifically connects MT and V1—the tract by which the purported backprojections would presumably propagate. Finally, researchers could set up a multi-array setup of low intensity focused ultrasound pulsation (LIFUP) devices—one targeting MT, one targeting V1, and one focused on the unveiled white-matter tract between them. Since LIFUP has a near three-fold increase in neuromdulatory spatial specificty compared to TMS (Bystritsky, 2011), I would feel more confident in the actual, selective modulation of the intended areas. By using LIFUP at frequencies that encourage neural activity, one could replicate Marg (1994) and Hotson’s (1994) TMS findings after stimulation to V1 and V5. Furthermore, LIFUP could be used to selectively depress the propagation of activity in the elucidated white matter between V1 and V5—rendering communication between the two regions ineffective. This affords the experimental opportunity to activate both V1 and V5 alongside the depression of their white matter connectivity in a temporally synchronous manner.
Thus, if a subject reported seeing a moving phosphene under this stimulation condition (a naturally implausible state of affairs without the advent of such neuromodulatory technology) it would be evidence that it is merely the temporally synchronous activity of specialized cortical regions that gives rise to conscious percepts of visual motion independent of their back-and-forth communications. If subjects do not report this moving phosphene, the study would conclude by exactly replicating the Pascual-Leaon (2001), but with the increased spatial resolution afforded by LIFUP. This latter course of events, pending a replication, would permit the default on accepting the working theory that V5’s recurrent connectivity with V1 is necessary for the formation of a unified conscious percept.
Bullier, Jean. “Feedback connections and conscious vision.” Trends in cognitive sciences 5.9 (2001): 369-370.
Bullier, Jean, et al. “The role of feedback connections in shaping the responses of visual cortical neurons.” Progress in brain research 134 (2001): 193-204.
Bullier, J. What is fed back? In Twenty-three Questions for the 21st Century (Sejnowki, T. and Van Hemmen, L., eds), Oxford University Press
Bystritsky, Alexander, et al. “A review of low-intensity focused ultrasound pulsation.” Brain stimulation 4.3 (2011): 125-136.
Descoteaux, Maxime, et al. “Deterministic and probabilistic tractography based on complex fibre orientation distributions.” Medical Imaging, IEEE Transactions on 28.2 (2009): 269-286.
Fuggetta, Giorgio, Antonio Fiaschi, and Paolo Manganotti. “Modulation of cortical oscillatory activities induced by varying single-pulse transcranial magnetic stimulation intensity over the left primary motor area: a combined EEG and TMS study.” Neuroimage 27.4 (2005): 896-908.
Heller, Joshua, et al. “Information flow and temporal coding in primate pattern vision.” Journal of computational neuroscience 2.3 (1995): 175-193.
Hotson, John, et al. “Transcranial magnetic stimulation of extrastriate cortex degrades human motion direction discrimination.” Vision research 34.16 (1994): 2115-2123.
Lamme, Victor AF, and Pieter R. Roelfsema. “The distinct modes of vision offered by feedforward and recurrent processing.” Trends in neurosciences 23.11 (2000): 571-579.
Marg, Elwin, and David Rudiak. “Phosphenes induced by magnetic stimulation over the occipital brain: description and probable site of stimulation.” Optometry & Vision Science 71.5 (1994): 301-311.
Nowak, Lionel G., and Jean Bullier. “The timing of information transfer in the visual system.” Extrastriate cortex in primates. Springer US, 1997. 205-241.
Nowak, L. G., et al. “Visual latencies in areas V1 and V2 of the macaque monkey.” Visual neuroscience 12.02 (1995): 371-384.
Panizza. Osservazioni sul nervo ottico. Gior. I. R. Ist Lomb. Sci. Lett. Arti., 7 (1855), pp. 237–252
Pascual-Leone, Alvaro, and Vincent Walsh. “Fast backprojections from the motion to the primary visual area necessary for visual awareness.” Science 292.5516 (2001): 510-512.
Raiguel, Steven E., et al. “Response latencies of visual cells in macaque areas V1, V2 and V5.” Brain research 493.1 (1989): 155-159.
Zeki, Semir, et al. “A direct demonstration of functional specialization in human visual cortex.” The Journal of neuroscience 11.3 (1991): 641-649.
Zeki, Semir, and D. H. Ffytche. “The Riddoch syndrome: insights into the neurobiology of conscious vision.” Brain 121.1 (1998): 25-45.