The Signs of Silence

Abstract

How does auditory cortical respond to silence? Fukushima and colleagues show that absent external input, activity in macaque auditory cortex is still highly structured. This structure likely reflects adaptive tuning that mates auditory analysis with effective action and perception.

Spontaneous brain activity has puzzled and intrigued neuroscientists since it became possible to routinely monitor the electroencephalogram (EEG) using noninvasive electrical recordings from the human scalp. Nonetheless, neuroscience investigations have generally shied away from spontaneous activity in favor of sensory responses or motor-related activity, because it is relatively easier to align one's analytic strategy with events that can be objectively and accurately measured, such as a sensory stimulus onset or a motor response. Recent technological, analytic and conceptual developments have led to a resurgence of interest in spontaneous activity (Raichle, 2010), however, a conceptual problem remains. On one hand, it seems obvious that spontaneous activity reflects what the brain is doing at the moment – recovering from stimulus processing or behavioral responding, preparing for expected inputs, or an upcoming behavioral response, maintaining items in working memory, vegetative functions, etc. On the other hand, it is seldom clear exactly which of these activities, or which combination of them is in play in a given moment, and thus, many prefer less pejorative terms like ‘ongoing,’ or ‘ambient,’ or ‘pre-stimulus’ activity. In any case, ongoing arguably ‘spontaneous’ activity accounts for the majority of brain energy utilization (Raichle, 2010), and has a complex dynamic structure spanning the frequency spectrum, as illustrated by cross frequency coupling, measured both within and across locations [rev by (Canolty and Knight, 2010)]. Furthermore, ongoing ‘prestimulus’ activity demonstrably affects stimulus processing and behavioral responding (Lakatos et al., 2008;Womelsdorf et al., 2006), and likely underpins consciousness (DeHaene and Changeux, 2011).

The paper by Fukushima and colleagues in this issue of Neuron takes this theme in an important direction – the manner in which the structural and functional organization of a brain region is mirrored in its ambient activity. Specifically, this team investigated the idea that structured spontaneous activity in the macaque auditory cortex has a systematic relationship to underlying organizational features, such as the rostral-to-caudal gradient in the pure-tone frequency preferences of neurons and mirror image reversals in this gradient that occur at boundaries between cortical areas. Fukushima and colleagues used micro-ElectroCorticoGraphy (micro-ECoG) recorded from dense electrode arrays (1 mm spacing) placed directly on the pial surface of the cortex to map and compare ongoing (spontaneous) activity with tone-evoked responses from regions along the supratemporal plane extending forward from primary auditory cortex (A1). They conclude that there is indeed a close relationship between functional organization and spontaneous activity in the auditory cortical regions of the supratemporal plane. This use of the micro-ECoG method is an innovative and potentially important approach, and raises a number of implications as well as underscoring important open questions.

Methodologically, the paper showcases the strengths of micro-ECoG in providing a wide-range view of functional organization in a large cortical network including core auditory cortices, A1 and the rostral area (R), as well as more anterior regions. As pointed out by the authors, the view they provide is on a scale comparable to that provided by previous fMRI studies in both human and nonhuman primates. Critically, micro-ECOG yields this view with high temporal resolution, utilizing amplitude fluctuations in a specific range of the neuronal activity spectrum, high-gamma (60-250 Hz). The amplitude of neuronal activity in the high-gamma frequency range provides a relatively uncomplicated index of massed firing in neuronal ensembles underlying the electrodes, as well as a relatively direct linkage to studies using this exact measurement for studying brain activity in surgical epilepsy patients (Canolty and Knight, 2010).

Fukushima and colleagues were able to use micro-ECoG to detail a relationship of spontaneous activity to functional architecture. Specifically, they verified that the high-gamma fraction of the stimulus-evoked response can be used to outline tonotopic maps within core and more rostral areas, and the mirror-symmetric reversals at area boundaries as demonstrated by a host of earlier studies. They then cross-registered tonotopic maps with maps derived from spontaneous activity using the same high-gamma measure. This is a large step forward, as it begins to bridge the gap between a reasonably well-evolved understanding of how auditory cortex responds to stimulus input with the deeper issues surrounding ongoing activity, and all the neuronal activities that compete and/or collaborate in this period, as discussed above. The fact that the rules governing ongoing neuronal activity are – at least to some extent – determined by the structural and functional organization of a given brain region highlights the need for a better understanding of the underlying neuronal circuitry.

Fukushima and colleagues relate their findings to several current questions in systems neuroscience, of which we highlight two here. One key issue they discuss is the impact of ongoing activity on stimulus processing; a variety of findings indicate that ongoing fluctuations of activity have a large impact on the parameters of stimulus-evoked responses, stimulus detection and the efficiency of behavioral responding. To be clear, these ‘activity fluctuations’ usually reflect synchronous, rhythmic excitability variations (oscillations) in interconnected ensembles of local neurons (Jensen et al., 2012;Schroeder and Lakatos, 2009). We will elaborate this theme below. Another important current issue is the specific physiological interpretation of high gamma power fluctuation. Fukushima et al. are inclined toward the position that the signal arises primarily from neuronal spiking in the superficial layers of auditory cortex, based on a proximity argument, and on a prior study in rodent auditory cortex. This seems to us unlikely given that in the auditory cortices of the awake monkey, the massive weight of both stimulus-evoked and spontaneous firing is in the granular layers compared to the relatively sparse firing seen in the more superficial layers [see e.g., (Kajikawa and Schroeder, 2011)]. Assuming as the authors do, that high-gamma power is related to multiunit firing, high-gamma generated by high volume firing in the middle layers is likely to overwhelm any generated by the much more sparse firing in supragranular sites.

CAVEATS AND OPEN QUESTIONS FOR FUTURE RESEARCH

Fukushima and colleagues raise a number of logical possibilities regarding underlying causes of structure in ongoing auditory cortical activity, based on a detailed consideration of the relevant anatomical connectivity patterns between core and higher order cortices, and between auditory core and thalamic regions. They also discuss a provocative idea that ongoing activity in auditory cortex represents a playback of recently experienced stimulation. Continuing down this path to longer time scales, it is noteworthy that the dynamical structure of spontaneous activity across the spectrum in auditory cortex bears a remarkable, and likely non-coincidental, resemblance to the 1/f statistics of the natural auditory environment (Garcia-Lazaro et al., 2006). This fits with the idea that the blueprint for macaque auditory cortex evolved under the pressures of this natural environment, and that in ontogeny, individuals’ auditory cortices further tune to the statistics of that same environment (Berkes et al., 2011). It will be interesting to investigate these relationships further and to see how nature and nurture collaborate in this arena.

Needless to say, the causes of ‘spontaneous order’ in auditory as well as other cortices is a prime area for future research, as currently there are many more questions than answers. For example, the authors note work by Raichle and colleagues on so-called ‘resting state’ fMRI as evidence that the brain is constantly active, a line of work that has virtually exploded as a means of mapping large scale brain functional connectivity networks using graph theoretic analyses (Bullmore and Sporns, 2009). To connect the dots, it is interesting to note that this approach is in principle applicable at smaller scales such as those dealt with here, which would in effect, represent subsets or nodes in a larger network. This in turn underscores the point (see also below) that it will be important to relate high-gamma to lower frequency dynamics, extending down to the infraslow ranges that approximate the timeframe of hemodynamic oscillations. Along these lines, it is important to note that for low frequency oscillations, phase rather than amplitude, carries information about the auditory environment (Kayser et al., 2009), and there is indeed indication that tonotopic maps might be at least partially reflected in the phase of ongoing low frequency oscillations modulated by auditory input (O'Connell et al., 2011) . It will be interesting to see whether such phase-maps occur spontaneously, like the high gamma amplitude maps described by Fukushima et al.

Based on the above considerations, the focus on high gamma power is reasonably justified in this context, but the findings of Fukushima el al. should not be taken to indicate that this measure gives a readout of cortical activity that is superior to that provided by lower frequency measures. This is particularly the case when it comes to analyzing the brain's representation of complex, natural stimulus patterns and movements. It remains likely that analyzing lower as well as high-gamma frequencies, albeit more complicated, will provide the best readout of the information the brain has encoded (Kayser et al., 2009). As Fukushima and colleagues note, spontaneous activity displays a great deal of cross frequency coupling, wherein the phase of lower frequency regulates amplitude in higher frequencies (e.g., high gamma), as well as associated variations in neuronal firing. The variation in the strength of cross-frequency coupling – like low frequency phase and high frequency amplitude – might provide an additional useful measure of neuronal activity, both within and across different nodes of sensory processing (Canolty and Knight, 2010), since these relationships across frequencies appear important in parsing and integrating information along the sensory processing hierarchy (Buzsaki, 2010).

The manner in which spontaneous activity reflects the current state of the system is an issue dealt with at length by Fukushima and colleagues. They make a number of excellent points including the fact that when regularities in the ongoing stimulus context permit the brain to make predictions about upcoming stimulus timing, the re-arrangement of ongoing activity in auditory cortex can make an instrumental contribution to effective stimulus processing, molding it to the current goals of the observer. As noted, this is a hot topic in systems neuroscience, and to it we would add that in constructing experiments and interpreting findings, it will be critical to consider the mode in which the system is operating in order to meet task demands (Schroeder and Lakatos, 2009). That is, are there regularities that allow the brain to make predictions, such as in listening to speech, or to the sounds of a person walking past us, or are task-relevant stimuli emerging randomly (temporally unpredictable), as in a cat watching a mouse hole or a taxi cab driver waiting for a traffic light to change? In the first (temporally predictable) case, ongoing lower frequency activity can imbue the brain's predictions; i.e., it can entrain to the rate of the events in the stream so that high excitability states coincide with these events, thus enhancing their representation. In the latter (random) case, ongoing activity likely will reflect suppression of slower excitability fluctuations and up-regulation of sensitivity to prepare for events whose timing is unpredictable. In both cases, the structure of ongoing activity, or the internal, neurophysiological context (Buzsaki and Chrobak, 1995) is modulated to best meet task demands, in order to most efficiently process the relevant content. We heartily second the authors’ comment that it will be important to see how the top down modulation of ongoing activity is superimposed on the structure-dependent modulations they describe.