Cortical feedback depolarization waves: A mechanism of top-down influence on early visual areas

Abstract

Despite the lack of direct evidence, it is generally believed that top-down signals are mediated by the abundant feedback connections from higher-to lower-order sensory areas. Here we provide direct evidence for a top-down mechanism. We stained the visual cortex of the ferret with a voltage-sensitive dye and presented a short-duration contrast square. This elicited an initial feedforward and lateral spreading depolarization at the square representation in areas 17 and 18. After a delay, a broad feedback wave (FBW) of neuron peak depolarization traveled from areas 21 and 19 toward areas 18 and 17. In areas 18 and 17, the FBW contributed the peak depolarization of dendrites of the neurons representing the square, after which the neurons decreased their depolarization and firing. Thereafter, the peak depolarization surrounded the figure representation over most of areas 17 and 18 representing the background. Thus, the FBW is an example of a well behaved long-range communication from higher-order visual areas to areas 18 and 17, collectively addressing very large populations of neurons representing the visual scene. Through local interaction with feedforward and lateral spreading depolarization, the FBW differentially activates neurons representing the object and neurons representing the background.T he current view of perception and cognition is that they rely on three main brain mechanisms, each supported by the existence of particular anatomical connections: bottom up, i.e., processing by early sensory areas, which is conveyed to higherorder areas; lateral processing through horizontal connections within an area; and top-down modulatory influences exerted by the rather extensive anatomical connections from higher-order sensory areas to the cortex in early sensory areas. Despite the fact that these top-down connections have been known for Ͼ25 years (1), and despite an overwhelming number of reports in which one could interpret the observations as presumed effects of top-down modulations, there is still no direct physiological evidence revealing the mechanism(s) by which higher-order sensory areas alter the computations of neurons in early sensory areas (2). That is, there is no evidence how, when, and where the top-down inputs alter the computation of neurons in early sensory areas. Further, the relative importance and timing of local lateral computations and top-down effects in previous studies of object perception are not obvious (3-5).Here, we define top-down modulation as a mechanism by which higher-order sensory areas through their connections influence computations of neurons in early sensory areas. These connections typically target neurons in upper (supragranular) layers or in lower (infragranular) layers within these early areas. Theoretically, it has been proposed that lateral interactions and the eventual feedback from higher-order visual areas would be finely timed to engage a large population of supragranular neurons in areas 17 and 18 in a cooperative computation of the visual stimulus and its surrou...

Characteristic sounds facilitate visual search

“…Once the high-level visual representation of the target is enhanced by the associated auditory signal, localization of the target could be facilitated via the extensive crossconnections between the ventral (thought to mediate object processing) and dorsal (thought to mediate spatial and action-related processing) cortical visual pathways (e.g., Felleman & Van Essen, 1991), and/or via the feedback connections from high-level polysensory areas and object-processing visual areas to low-level retinotopic visual areas (e.g., Rockland & Van Hoesen, 1994;Roland et al, 2006). For example, within about 100 msec after presentation of a visual stimulus, a wave of feedback activation from high-level visual areas selectively enhances the low-level retinotopic (spatially selective) responses to the stimulus (Roland et al, 2006).…”

“…For example, within about 100 msec after presentation of a visual stimulus, a wave of feedback activation from high-level visual areas selectively enhances the low-level retinotopic (spatially selective) responses to the stimulus (Roland et al, 2006). It is plausible that when the high-level object-based visual responses to the target (e.g., a cat) are cross-modally enhanced by the simultaneously presented characteristic sound (e.g., a meow), those enhanced high-level visual responses in turn strengthen the feedback enhancement of retinotopic responses to the target stimulus in low-level visual areas.…”

Voltage-sensitive dye imaging of transcranial magnetic stimulation-induced intracortical dynamics

“…In contrast to imaging methods applicable in humans, this method is invasive but allows avoiding the commonly experienced contamination of signals by artifacts due to the strong TMS-induced electric field. In combination with a tandem-lens system of large numerical aperture (23) and a fast CCD camera as detector, VSD imaging captures several square millimeters of cortex with an emphasis on superficial layers (24)(25)(26)(27)(28)(29)(30)(31)(32), allowing us to record activity changes within milliseconds across millions of neurons at once with a spatial resolution of ∼50 μm (for review see ref. 33).…”

Cortico-cortical Communication Dynamics

“…Part of the explanation might be that the dye signal in vivo reflects synaptic activity at the mesoscopic scale, whereas the action potential recordings capture the activity of single neurons (Lippert et al, 2007;Eriksson et al, 2008). Nevertheless, in several studies one can follow how net increases in the synaptic activity propagate over the cortical areas when the cortex is perturbed by a sensory transient (Senseman, 1996;Prechtl et al, 1997;Senseman and Robbins, 2002;Slovin et al, 2002;Grinvald and Hildseheim, 2004;Roland et al, 2006;Ferezou et al, 2007;Lippert et al, 2007;Xu et al, 2007;Ahmed et al, 2008;Han et al, 2008;Takagaki et al, 2008;Yoshida et al, 2008;Harvey et al, 2009;Ayzenshtat et al, 2010;Meirovithz et al, 2010;Ng et al, 2010;Polack and Contreras, 2012;Harvey and Roland, 2013). This synaptic dynamics may show some order in the feed-forward propagation of net-excitation for example between V1 and V2 in monkeys, rats and turtles, between the barrel field and the motor cortex in the mouse, and between visual areas 17, 18 and 19, 21 in the ferret.…”

“…The whole primary auditory cortex was engaged in 26-40 ms after stimulus start in guinea pigs (Horikawa et al, 1998;Kubota et al, 2012). The whole craniotomy exposed part of the primary visual cortex in ferrets, cats, and monkeys became engaged 48-70 ms after stimulus start, even with small stimuli (Slovin et al, 2002;Jancke et al, 2004;Eriksson and Roland, 2006;Roland et al, 2006;Sharon et al, 2007;Eriksson et al, 2008;Harvey et al, 2009;Ayzenshtat et al, 2010;Meirovithz et al, 2010;Roland, 2010;Chavane et al, 2011;Reynaud et al, 2012;Harvey and Roland, 2013). In mice and rats it took some 70-110 ms for the dynamics to engage the whole primary visual cortex Han et al, 2008;Gao et al, 2012;but Lim et al, 2012: 46 ms;Polack and Contreras, 2012).…”