Neuron

Perspective

Rhythms for Cognition:

Communication through Coherence
Pascal Fries1,2,*
1Ernst Strüngmann Institute (ESI) for Neuroscience in Cooperation with Max Planck Society, 60528 Frankfurt, Germany
2Donders Institute for Brain, Cognition and Behaviour, Radboud University Nijmegen, 6525 EN Nijmegen, Netherlands
*Correspondence: [email protected]
http://dx.doi.org/10.1016/j.neuron.2015.09.034

I propose that synchronization affects communication between neuronal groups. Gamma-band (30–90 Hz)
synchronization modulates excitation rapidly enough that it escapes the following inhibition and activates
postsynaptic neurons effectively. Synchronization also ensures that a presynaptic activation pattern arrives
at postsynaptic neurons in a temporally coordinated manner. At a postsynaptic neuron, multiple presynaptic
groups converge, e.g., representing different stimuli. If a stimulus is selected by attention, its neuronal rep-
resentation shows stronger and higher-frequency gamma-band synchronization. Thereby, the attended
stimulus representation selectively entrains postsynaptic neurons. The entrainment creates sequences of
short excitation and longer inhibition that are coordinated between pre- and postsynaptic groups to transmit
the attended representation and shut out competing inputs. The predominantly bottom-up-directed gamma-
band influences are controlled by predominantly top-down-directed alpha-beta-band (8–20 Hz) influences.
Attention itself samples stimuli at a 7–8 Hz theta rhythm. Thus, several rhythms and their interplay render
neuronal communication effective, precise, and selective.

Imagine a model of the human brain that is both complete to the in (ultra)structural connectivity, neuronal synchronization as
point of producing behavior that is indistinguishable from human an emergent dynamic of active neuronal groups has causal
behavior and detailed to the point of atomistic resolution. This consequences for neuronal communication. If neuronal commu-
hypothetical model would be an invaluable tool in place of imper- nication depends on neuronal synchronization, then dynamic
fect experimental recordings from living subjects by providing changes in synchronization can flexibly alter the pattern of
complete downloads from the model. However, those down- communication. Such flexible changes in the brain’s communi-
loaded data would require analysis and interpretation, just like cation structure, on the backbone of the more rigid anatomical
experimental data, before any scientific insight were achieved. structure, are at the heart of cognition.
Scientific insight is human insight, and human insight into the
brain proceeds just as human insight into anything else out there Communication through Coherence
in the world. The world provides a wealth of sensory data, in Because the main thrust of the concept is that neuronal com-
which regularities, relations, and rules need to be found to arrive munication is subserved by neuronal synchronization, often
at an understanding of the perceived processes. Such an under- quantified by the coherence metric, I have named the concept
standing may be referred to as a mental model, which restricts ‘‘Communication through Coherence,’’ or CTC. I formulated
itself parsimoniously to the aspects crucial for capturing the the CTC hypothesis 10 years ago (Fries, 2005) and aim here to
essence of the perceived. It might be characterized as abstract provide a revised formulation of CTC that takes into account
and semantic, and it is certainly incomplete in the sense that it the plethora of CTC-relevant data, which has been generated
discards the rich initial data for an intuitive or conceptual under- in the meantime, and that further distils the essence of CTC.
standing of generative principles behind the data. To present The Essence of CTC
such a concept of neuronal processing, I will define as ‘‘neuronal Here are the essential propositions of the CTC hypothesis: An
representation’’ the spatial activation pattern in a group of neu- activated neuronal group tends to engage in rhythmic synchroni-
rons; as ‘‘neuronal communication’’ the transfer of one represen- zation. Rhythmic synchronization creates sequences of excita-
tation in a presynaptic, or sending, group to a new representation tion and inhibition that focus both spike output and sensitivity
in a postsynaptic, or receiving, group; and as ‘‘neuronal compu- to synaptic input to short temporal windows. The rhythmic
tation’’ the transformation that happens between the represen- modulation of postsynaptic excitability constitutes rhythmic
tations. This illustrates the central role of communication as modulations in synaptic input gain. Inputs that consistently
the process that implements computation and thereby creates arrive at moments of high input gain benefit from enhanced
new representations. effective connectivity. Thus, strong effective connectivity re-
Neuronal communication has classically been conceived of as quires rhythmic synchronization within pre- and postsynaptic
being determined by structural anatomical connectivity and by groups and coherence between them, or in short—communica-
activity-dependent changes to the anatomical (ultra)structure tion requires coherence. In the absence of coherence, inputs
of the connection. I propose that even in the absence of changes arrive at random phases of the excitability cycle and will have

220 Neuron 88, October 7, 2015 ª2015 Elsevier Inc.

Neuron

Perspective

a lower effective connectivity. A postsynaptic neuronal group groups in a lower visual area compete to communicate with
receiving inputs from several different presynaptic groups re- one target group in a higher visual area. These studies will be
sponds primarily to the presynaptic group to which it is coherent. discussed in detail below.
Thereby, selective communication is implemented through se- Challenges for the Original CTC Hypothesis
lective coherence. While there is substantial experimental support, some studies
The fundamental proposition, that a postsynaptic rhythm posed challenges to the original CTC formulation that motivated
modulates input gain, has received direct experimental support. the new CTC formulation. If the two communicating neuronal
When optogenetic stimulation is used to drive fast-spiking groups are bidirectionally coupled, I originally proposed zero-
interneurons in somatosensory cortex with a 40 Hz pulse train, phase synchronization (see Figure 3 of Fries, 2005). To cope
the network resonates at gamma frequency, and the precise with increasing delays for increasingly distant groups, I consid-
timing of vibrissa deflections relative to the pulse train affects ered lower frequencies. However, recent studies have demon-
both neuronal (Cardin et al., 2009) and behavioral (Siegle et al., strated that neuronal groups in widely separated areas can be
2014) responses. Similarly, when a weak sustained muscle coherent in the gamma band, i.e., at a relatively high frequency
contraction induces a corticospinal beta rhythm, transcranial (Bastos et al., 2015a; Bosman et al., 2012; Gregoriou et al.,
magnetic stimulation pulses applied to motor cortex lead to 2009). At the same time, it became clear that even though the
muscle responses that depend on the phase of stimulation in areas are bidirectionally coupled, this gamma-band coherence
the beta rhythm (van Elswijk et al., 2010). Note that the transmis- does not occur at zero phase, but with a systematic delay, i.e.,
sion of spikes from one area to the next does not only depend on with a directedness (Bastos et al., 2015b; Bosman et al., 2012;
coherence between the areas but also on synchronization within Gregoriou et al., 2009; Grothe et al., 2012; Jia et al., 2013a; Zand-
the sending area, as has been demonstrated, e.g., between vakili and Kohn, 2015). I had originally suggested such delayed
areas V1 and V2 (Jia et al., 2013a; Zandvakili and Kohn, 2015). coherence, consistent with a directed entrainment, if the presyn-
The subsequent proposition, that strong effective connectivity aptic group projects unidirectionally to the postsynaptic group.
requires coherence between pre- and postsynaptic groups, has In the new CTC formulation, I suggest that unidirectional entrain-
also been supported. One study investigated the relation be- ment occurs separately in both directions of a bidirectional
tween the effective connectivity and the phase relation for pairs communication link, as I will explain in more detail below.
of recording sites in visual cortex of awake cats and monkeys A second challenge to the original CTC came from the fact that
(Womelsdorf et al., 2007). For each trial, the phase relation be- it had not specified precisely how postsynaptic excitability varies
tween gamma rhythms at the two recording sites was deter- with phase. Mathematical implementations of CTC that assume
mined, and trials were sorted accordingly into phase-relation a sinusoidal oscillation and a linear relation between phase and
bins. Across all trials within a phase-relation bin, effective excitability show that presynaptic groups, which are incoherent
connectivity was then determined. This showed that effective to the postsynaptic group, might still have a substantial impact
connectivity depends on the phase relation. Effective connectiv- (Akam and Kullmann, 2012). The new CTC proposes that excit-
ity is maximal for the phase relation at which the two sites ability is modulated by rhythmic synchronization in a way that
typically synchronize. Phase relations supporting interactions is neither sinusoidal nor linear, in agreement with mathematical
between the groups precede those interactions by a few milli- models entailing spiking excitatory and inhibitory neurons (Börg-
seconds, consistent with a mechanistic role. ers and Kopell, 2008; Cannon et al., 2014; Gielen et al., 2010).
The final proposition, that selective communication is imple- The New CTC
mented through selective coherence, has received experimental I will illustrate the new CTC, from here on mostly referred to just
support from studies using selective visual attention or selective as CTC, for the case of visual cortical gamma-band synchroni-
movement intention. One study tested this prediction in the zation. When visual cortex of an awake and attentive subject is
human motor system, assessing activity from bilateral motor activated by an appropriate stimulus, the activated neurons
cortices with magnetoencephalography and corresponding spi- engage in rhythmic synchronization in the gamma-frequency
nal activity through electromyography of bilateral hand muscles band (30–90 Hz) (Fries et al., 2001; Gray et al., 1989). This holds
(Schoffelen et al., 2011). During bimanual wrist extension, each during natural viewing (Figure 1; Boxes 1 and 2) (Brunet et al.,
motor cortex showed coherence with its contralateral hand 2015). During the gamma cycle, excitatory neurons trigger local
muscle. One of the hands was cued as the response hand to inhibitory neurons within about 3 ms (Buzsáki and Wang, 2012;
report an unpredictable visual go cue. The corticospinal connec- Csicsvari et al., 2003; Fries et al., 2007; Hasenstaub et al.,
tion that effectuated the subsequent motor response showed 2005; Salkoff et al., 2015; Vinck et al., 2013). When the ensuing
enhanced corticomuscular coherence in the gamma-band (40– inhibition of the local network decays, the gamma cycle starts
47 Hz). This effect was observed in the absence of changes in again with a new round of excitatory neuron spiking. Thus, there
motor output or changes in local cortical gamma-band synchro- is only a 3 ms window for excitation, whereas the longer rest of
nization. Thus, selective movement intention is implemented the gamma cycle is dominated by inhibition (Figure 2A). The
by selective gamma-band coherence. Enhanced corticospinal spikes travel from the presynaptic neurons, through their
gamma-band coherence during movement preparation corre- anatomical projections, to the postsynaptic neurons, where
lates closely with shortened reaction times (Schoffelen et al., they trigger excitatory neuron spiking followed by inhibitory
2005). Yet probably the most compelling evidence for selective neuron spiking. The ensuing inhibition essentially closes the
communication through selective coherence comes from door in front of other inputs, because it strongly reduces their
studies of selective visual attention, during which two neuronal synaptic input gain.

Neuron 88, October 7, 2015 ª2015 Elsevier Inc. 221

 Neuron

Perspective
Figure 1. Natural Viewing Induces a
Gamma-Band Rhythm in Visual Cortex
Raw LFP trace recorded as the voltage between
two neighboring electrocorticographic electrodes
on primary visual cortex of an awake macaque
monkey during one visual exploration of the
photograph of two oranges (adapted and modified
from Brunet et al., 2015). Green vertical line in-
dicates stimulus onset, red vertical lines indicate
saccades. Insets show the stimulus and super-
imposed the eye position trace around that time
point in blue, and the eye position trace during
this exploration so far in gray. Prior to stimulus
appearance and free viewing, the monkey fixated
on a small central dot.

The unequal duty cycle of short excitation followed by longer challenge arising from observed interareal delays. In addition,
inhibition leads to a nonsinusoidal gain modulation (Figure 2B), systematic investigations of directed interareal influences as a
and the nature of inhibition, involving perisomatic shunting function of frequency revealed that influences in the two direc-
inhibition (Buzsáki and Wang, 2012), leads to a nonlinear gain tions predominate in distinct frequency bands (Bastos et al.,
modulation. If time constants are similar between entrained 2015a), as explained in detail below.
pre- and postsynaptic networks, the next round of synaptic in- The new CTC addresses challenges to the original CTC, but
puts from the presynaptic network will be timed to the moment one of the appealing predictions of the original CTC appears at
when inhibition in the postsynaptic network decays. This is first to be lost with this revision. If, as originally assumed, delays
because local excitation triggers local inhibition and thereby between sending and receiving groups are negligible, coherence
starts a few-millisecond timer corresponding to the inhibitory increases effective connectivity between the coherent groups in
time constant. At the postsynaptic network, excitatory inputs both directions. That is, feedback from the receiving group is
arrive with a delay, which starts the local timer in the postsyn- likely to be more effective at the coherent sending group than
aptic network with a corresponding delay. With precisely the at the noncoherent sending group, even if it is anatomically
same delay, the next round of excitation arrives. This entrain- directed to both (Fries, 2005). Thus, CTC might render anatom-
ment of the postsynaptic group to the rhythmic input from the ically nonselective feedback functionally selective to the appro-
presynaptic group automatically sets up a phase relation that priate sending group. If, as we now point out, delays are not
is optimal for CTC, as has been highlighted in mathematical negligible, both feedforward and feedback signaling will incur
models (Börgers and Kopell, 2008; Cannon et al., 2014; Gielen some delay. In the sending cortical column, re-entrant feedback
et al., 2010). inputs will arrive several milliseconds after feedforward outputs
In the new CTC formulation, I propose that entrainment with have been sent. Those re-entrant feedback inputs can only be
delay is the general mechanism that sets up phase relations coherent with a delayed version of the original output. Intrigu-
subserving CTC, both for unidirectional communication and for ingly, such delayed versions have been described. Single neu-
bidirectional communication. Bidirectional communication is rons in supragranular layers of macaque area V1 show visually
implemented separately for the two directions, via unidirectional induced gamma-band synchronization with a systematic delay
entrainment per direction. Anatomical data actually show that for of approximately 1 ms per 100 micron shift toward the cortical
each direction of communication, the communicating brain surface (Livingstone, 1996). Recently, laminar current-source-
areas have specialized neuronal groups, i.e., a given brain area density analysis revealed that this gamma delay extends from
has neurons receiving inputs and different neurons sending out- layer 4 toward both supra- and infragranular layers (Figure 3A)
puts (Felleman and Van Essen, 1991; Markov et al., 2014). Thus, (van Kerkoerle et al., 2014). Thus, interlaminar delays might delay
the new CTC takes anatomical data on interareal projections the supra- and infragranular gamma phase such that reentrant
more closely into account and resolves the abovementioned feedback arrives at the excitable phase of the same, delayed,

Box 1. Current Status of the Field

d Rhythmic synchronization, including gamma-band synchronization, is widespread across the nervous system and across
species.
d Different rhythms coexist and are often synchronized to each other or nested into each other.
d Gamma and beta rhythms modulate input gain, and their coherence subserves effective connectivity.
d Rhythmic synchronization in different frequency bands is highly structured across areas, layers, and the corresponding
projections.
d Patterns of synchronization change dynamically with stimulation and behavioral context in a way that strongly suggests that
selective coherence implements selective communication.

222 Neuron 88, October 7, 2015 ª2015 Elsevier Inc.

Neuron

Perspective

Box 2. Future Directions

d Does rhythmic synchronization, in particular gamma-band synchronization, occur generally in active neuronal groups, also un-
der fully natural conditions?
d Can we experimentally manipulate neuronal synchronization, while leaving other aspects of neuronal activity unchanged, and
show effects on neuronal communication?
d Through which cell types and mechanisms do top-down influences affect the strength and frequency of gamma-band synchro-
nization?
d Cortical inhibition is very local, leading to local gamma rhythms; thalamic inhibitory neurons are collected and linked in the
thalamic reticular nucleus. Does this allow only one gamma across the thalamus, which needs to be entrained for cortical signal
propagation?

gamma cycle (Figure 3B) (Bastos et al., 2015b). Note that inter- devoid of it (Tiesinga et al., 2004). This time-sharing is achieved
areal anatomical projections have laminar origins and targets by the gamma rhythm, which focuses perisomatic inhibition in
that are consistent with the pattern of gamma flow suggested one part of the gamma cycle, and thereby leaves another part
in Figure 3 (Felleman and Van Essen, 1991; Markov et al., free for neurons to respond to excitatory input. This excitatory
2014). For pairs of areas that are hierarchically very distant, re- phase of the gamma cycle sees decaying inhibition and a tempo-
entrant feedback would arrive too late to hit a matching phase. rally focused rise in excitation (Hasenstaub et al., 2005; Salkoff
Intriguingly, we found that for those area pairs, gamma-band et al., 2015; Vinck et al., 2013). Such synchronized excitation in
influences exist essentially only in the feedforward direction, as turn generates rapid postsynaptic depolarization, ideal to trigger
discussed in more detail below (Bastos et al., 2015a, 2015b). spikes (Azouz and Gray, 2003).
In the following, I will argue that gamma-band coherence I would like to speculate that the rapid balancing of network
establishes a communication protocol that is effective, precise, excitation by inhibition reduces the postsynaptic impact of
and selective, with one section devoted to each of these three as- neurons, whose spike rate is modulated by the alpha rhythm.
pects. In two further sections, I will discuss the differential roles of The alpha rhythm is strong in cortex that is not activated by a
gamma and alpha-beta rhythms in feedforward versus feedback stimulus and/or not addressed by top-down influences like
signaling and the role of the theta rhythm in attentional sampling. attention (Fries et al., 2008; van Ede et al., 2011; Worden et al.,
2000). It has therefore been suggested that the alpha rhythm
Gamma-Band Coherence Renders Communication provides inhibition (Jensen and Mazaheri, 2010). Yet inhibition
Effective is involved in all classical EEG rhythms, including the gamma
The gamma rhythm results in a rapid succession of excitation rhythm. The crucial characteristic of alpha seems to be that it
and inhibition, and this is likely necessary to activate postsyn- blocks the communication of local activity to connected
aptic neurons. A typical cortical neuron receives inputs from neuronal groups (van Dijk et al., 2008; Zumer et al., 2014). I sug-
thousands of neighboring neurons. Among those input neurons, gest that the rapid balancing, which renders gamma ideal for
network excitation is faithfully tracked by network inhibition neuronal communication, likely renders alpha ideal to preclude
(Renart et al., 2010). Therefore, individual neurons typically communication. In the gamma cycle, excitation rises within
receive excitation and inhibition in a balanced way (Haider few milliseconds, fast enough to lead to postsynaptic depolari-
et al., 2006; Shu et al., 2003), including neurons in awake monkey zation before inhibition terminates it. By contrast, in the alpha
visual cortex during visual stimulation (Tan et al., 2014). Network cycle, excitation rises over the course of about 50 ms, while
excitation drives network inhibition with a small delay. During inhibition likely follows with the same 3 ms delay; in fact, the
gamma-band synchronization, this delay has been found to be spiking of putative excitatory and inhibitory V4 neurons does
around 3 ms (Atallah and Scanziani, 2009; Csicsvari et al., not show a significant phase shift in the alpha cycle (Vinck
2003; Fries et al., 2007; Hasenstaub et al., 2005; Vinck et al., et al., 2013). Thus, alpha might be slow enough, such that
2013). I suggest that the gamma rhythm generates fluctuations network excitation cannot escape network inhibition, which can-
in network excitation and inhibition that are sufficiently rapid, cels postsynaptic effects and renders local activity functionally
such that excitation can essentially temporally escape its ever invisible to remote projection targets. This might allow holding
chasing inhibition. The gamma rhythm concentrates inhibition local representations ‘‘on-stock’’ for flexible access by top-
to certain parts of the gamma cycle, which appears necessary down selection mechanisms.
to provide moments devoid of inhibition that allow postsynaptic
neurons to spike at all (Tiesinga et al., 2004). An important type of Gamma-Band Coherence Renders Communication
inhibition is exerted by parvalbumin-positive (PV) interneurons Precise
and is targeted to the soma and perisomatic region, where it Gamma-band coherence renders communication not only effec-
can powerfully antagonize and shunt dendritic excitation (Buz- tive but also precise. The gamma rhythm times the inhibition in the
sáki and Wang, 2012). If this powerful shunting arrived in a postsynaptic group to vanish just before another round of synap-
temporally unstructured manner, it would prevent postsynaptic tic inputs arrives, and it focuses those synaptic inputs to arrive
spiking at almost all times. Spiking is enabled by temporally simultaneously. This synchronous arrival of synaptic inputs is
focusing shunting to certain time periods, while leaving others an important component of a precise communication protocol.

Neuron 88, October 7, 2015 ª2015 Elsevier Inc. 223

 Neuron

Perspective

A Figure 2. Communication through Coherence

(A) Two presynaptic neuronal groups in a lower visual area
provide input to a postsynaptic neuronal group in a higher
visual area. The lower groups represent two visual stimuli, an
apple and a pear. In each neuronal group, network excitation
(red) triggers network inhibition (blue), which inhibits the local
network. When inhibition decays, excitation restarts the
gamma cycle. The gamma rhythm of the apple-representing
presynaptic group has entrained the gamma rhythm in the
postsynaptic group. Thereby, the apple-representing pre-
synaptic group can optimally transmit its representation,
whereas the pear-representing presynaptic group cannot
transmit its representation.
(B) A simplified illustration in which network excitation and
inhibition are combined into network excitability. Red vertical
lines indicate excitatory neuron spiking and blue vertical lines
inhibitory neuron spiking.

small fraction of the synchronization cycle. Pulsa-

B tile communication results in pulsatile computation
and a pulsatile postsynaptic neuronal representa-
tion. An ideal test case of a neuronal stimulus rep-
resentation is the orientation selectivity of primary
visual cortex. A recent study investigated single
neurons in awake monkey V1 with regard to the
orientation selectivity of their firing rate responses.
Orientation selectivity was calculated separately
for spikes occurring at different times in the gamma
cycle (Figure 4) (Womelsdorf et al., 2012). The
gamma cycle was subdivided into several phase
bins. Spikes occurring close to the gamma phase
to which spikes synchronized on average showed
stronger orientation selectivity in their spike rates
than spikes occurring at other times. These data
suggest that the neuronal representation of visual
stimulus orientation pulsates with the gamma
cycle. Of course, spike rates themselves pulsate
with the gamma cycle, and the study controlled
that the pulsating neuronal representation was
not a trivial consequence of pulsating spike counts.
Thus, these results argue against the often-prac-
The CTC protocol conveys a neuronal representation manifest in ticed distinction (and rivalry) between a rate code and a syn-
the spatial pattern of spike rates in the presynaptic neuronal chrony code, but rather for an integration of those schemes
group. This presynaptic spike rate pattern translates into a spatial (Ainsworth et al., 2012). Specifically, the results suggest a
activation pattern among the synaptic inputs to the postsynaptic gamma-rhythmic pulsatile rate code, in which the spatial pattern
group. The synaptic input pattern, multiplied by the pattern of of spike rates holds representations, yet only during short tem-
synaptic strengths, determines the level of postsynaptic depolar- poral windows in the gamma cycle.
ization and subsequent spike rate. The spike response should While gamma-band synchronization primarily aligns the spike
ideally be a precise function of the neuronal representation output of a neuronal group in time, this alignment contains further
conveyed by the active set of synaptic inputs. Synaptic currents fine temporal structure (Havenith et al., 2011; Vinck et al., 2010,
(at least the dominant AMPA- and GABAA-receptor-mediated 2013). In visual cortex, neurons spike earlier in the gamma cycle
currents) decay within a few milliseconds. If synaptic inputs when they are driven by stimuli closer to their preferred stimulus
were jittered, even by merely a few milliseconds, this would (Vinck et al., 2010). Thereby, in the cortical stimulus selectivity
substantially compromise the precision of the postsynaptic map, a given stimulus results in a systematic gamma wave of
response, i.e., the degree to which it is determined by the presyn- spiking, sweeping from the more to the less strongly activated
aptic spatial spike rate pattern. By decreasing such jitter, CTC columns. Postsynaptically, the synaptic inputs from more
mechanisms likely increase postsynaptic response precision. strongly activated neurons will arrive earlier and will thereby
Essentially, CTC renders interneuronal communication pulsa- have a larger influence before inhibition curtails further effects
tile, because communication happens only during a relatively (Börgers and Kopell, 2008; Cannon et al., 2014).

224 Neuron 88, October 7, 2015 ª2015 Elsevier Inc.

Neuron

Perspective

A A

Figure 4. The Gamma Cycle Implements Pulsatile Neuronal

Representations
(A–C) Awake monkey V1 single unit during visual stimulation with a drifting
grating (adapted and modified from Womelsdorf et al., 2012).
(A) Spike probability as function of the gamma phase in LFPs, which were
recorded simultaneously from separate nearby electrodes.
(B) The colored bar shows the partitioning of the gamma cycle into eight phase
bins containing equal numbers of spikes (aligned to the phase in the gamma
cycle to which spikes synchronized on average). Orientation tuning curves
calculated separately for the eight gamma phase bins show a strong modu-
lation of orientation selectivity with gamma phase, even though spike count
was equal.
(C) Orientation selectivity index (OSI) as function of the gamma phase bin, in
which the spikes occurred.

Figure 3. Interlaminar Delays Might Compensate for Interareal

Feedforward and Feedback Delays Gamma-Band Coherence Renders Communication
(A) Analysis of current source density (CSD) derived from laminar recordings in
awake monkey area V1 (adapted and modified from van Kerkoerle et al., 2014).
Selective
Laminar CSD was averaged relative to troughs in the gamma-filtered LFP from In addition to rendering communication effective and precise,
layer 4. The analysis reveals interlaminar gamma-band synchronization with coherence also renders communication selective. If one set
systematic delays as a function of distance from layer 4.
of synaptic inputs, constituting one neuronal representation,
(B) The interlaminar delays might delay the supra- and infragranular gamma
phase such that reentrant feedback arrives at the excitable phase of the same, succeeds in triggering postsynaptic excitation followed by inhi-
delayed, gamma cycle (Bastos et al., 2015b). Cortical depth is indicated by bition, this inhibition closes the door in front of other inputs.
approximate position of the different cortical layers, abbreviated as L1–L6. Those other inputs are then unable to transmit the neuronal
Arrows indicate the proposed flow of gamma-mediated signaling. Interareal
arrows conform to the known laminar pattern of feedforward and feedback representation that they constitute, and they are unable to
connections (Markov et al., 2014). Note that the CSD analyses do not reflect trigger inhibition themselves. Thereby, the winning set of
the finding that gamma-band spike-LFP coherence is stronger in superficial as synaptic inputs conquers the perisomatic inhibition in the post-
compared to deep layers (Buffalo et al., 2011).
synaptic neuronal group, entrains it to its own rhythm, and
thereby establishes a communication link that is selective or,

Neuron 88, October 7, 2015 ª2015 Elsevier Inc. 225

 Neuron

Perspective

in other words, exclusive (Börgers and Kopell, 2008; Gielen conveying the attended stimulus. By contrast, in the CTC sce-
et al., 2010). nario, attentional top-down control does not address the synap-
CTC Implements Attentional Selection ses signaling the attended stimulus, but rather it simply ad-
This selective communication link is an ideal candidate mecha- dresses the corresponding neurons in the lower area. The
nism for implementing selective attention, concretely the selec- rhythmic synchronization of those neurons manages to entrain
tive routing of attended sensory representations. When, e.g., postsynaptic neurons and thereby achieves the increase in input
one out of several visual stimuli is attended, because it is behav- gain at the postsynaptic neurons.
iorally relevant, the early visual cortical representation of this at- This central prediction of the CTC hypothesis has recently
tended stimulus is preferentially communicated to postsynaptic received direct support from two independent yet very similar
neuronal groups, at the expense of other, unattended, stimuli experimental studies (Bosman et al., 2012; Grothe et al., 2012).
(Reynolds et al., 1999). From lower to higher areas of visual cor- We illustrate here one of the studies (Figure 5) (Bosman et al.,
tex, neuronal projections converge such that postsynaptic 2012). Macaque monkeys fixated while two stimuli, positioned
neurons respond selectively to particular conjunctions of simpler next to each other at the same eccentricity a few degrees away
stimulus features; at the same time, this convergence renders from fixation, were presented either separately (Figures 5A–5D)
responses invariant, i.e., insensitive to stimulus dimensions like or simultaneously (Figures 5E–5H). During simultaneous presen-
the precise stimulus position, which are already represented tations, one of the stimuli was behaviorally relevant, the other
with high precision in early visual areas (Ito et al., 1995). This was irrelevant, and the monkey’s behavior indicated that the rele-
increasing spatial invariance with increasing hierarchical level vant stimulus was attended and the irrelevant one ignored.
is reflected in increasing receptive field (RF) sizes. Invariance ap- Neuronal recordings were performed simultaneously in the lower
pears necessary, because it readily offers a neuronal mechanism visual area V1 and the higher visual area V4. The two stimuli acti-
of stimulus recognition unperturbed by stimulus deviations in vated two separate groups of neurons in V1 (Figure 5B), and
irrelevant details. Also, it avoids a combinatorial explosion that there were neuronal groups in V4 that were activated by either
would result, if object-selective higher-area neurons represented stimulus to approximately the same degree (Figure 5D). When
particular sensory realizations of particular object tokens. Yet, one of the stimuli was presented separately, it induced a gamma
the convergence that produces both stimulus selectivity and rhythm in its respective V1 neuronal group, which entrained the
invariance unavoidably results in a situation in which a given V4 neuronal group (Figure 5C). When both stimuli were presented
postsynaptic neuron often receives synaptic inputs containing simultaneously, both stimuli induced a gamma rhythm in their
the representations of more than one perceptual object (Fries, respective V1 neuronal group (Figure 5F). Crucially, when one
2009). When a single neuron in a higher visual area responds of the two simultaneously presented stimuli was attended, only
with different firing rates to different stimuli, the simultaneous the corresponding V1 gamma managed to entrain the V4 gamma
presentation of both stimuli in the neuron’s RF results in a firing (Figure 5G). The ignored stimulus induced a gamma rhythm in V1,
rate that is a weighted average of the response to the isolated which did however fail to entrain the V4 gamma. The CTC
stimuli (Reynolds et al., 1999). This is the case when attention hypothesis suggests that this selective entrainment of V4 gamma
is directed away from both stimuli. However, when attention by the attended V1 gamma is the cause for the selective routing
is directed toward one of the two stimuli, the firing rate of the of the attended stimulus from V1 to V4. This proposal is also
postsynaptic neuron represents primarily the attended stimulus supported by corresponding mathematical models (Börgers
(Reynolds et al., 1999; Zhang et al., 2011). In the same attention and Kopell, 2008; Gielen et al., 2010).
tasks, presynaptic neurons in lower visual areas, whose smaller These results raise important questions for further investiga-
RFs contain only one of the two stimuli, show only small effects tion. In particular, we will need to understand better how atten-
of attention on their firing rates (Luck et al., 1997). Thus, with two tional top-down influences result in the selective interareal
visual stimuli, there are two sets of presynaptic neurons in lower synchronization. In the CTC scenario, attentional top-down influ-
visual areas, with firing rates hardly affected by attention, and ences do not need to address the synapses to postsynaptic neu-
postsynaptic neurons in higher visual areas, with firing rates rons in a higher area, where the competing stimuli are difficult
dominated by the attended stimulus. This can be modeled to disentangle. Rather, attentional top-down influences can be
elegantly if attention modulates the effective strength of the addressed to the presynaptic neurons in a lower area, which
synaptic inputs from lower visual neurons onto higher visual are arranged in maps, with neurons of similar preferences
neurons, i.e., the synaptic gain, and the respective models are located close to each other. Attentional top-down control can
explicitly ignorant about the mechanism of gain modulation therefore address neurons belonging to the neuronal representa-
(Reynolds et al., 1999; Reynolds and Heeger, 2009). I suggest tion of a given stimulus often by simply addressing a spatially
that CTC is an ideal mechanism to implement the attentional coherent neuronal group. This holds most straightforwardly for
modulation of input gain. As the term ‘‘input gain’’ conveys, the spatially specific attention. I would like to speculate that top-
attentional modulation acts on the sets of synaptic inputs down influences implementing attention to nonspatial stimulus
signaling the competing stimuli. At the postsynaptic neuron, features similarly address neurons in areas, where those stim-
those sets of synaptic inputs are likely distributed over the den- ulus features are represented in topographically ordered and
dritic tree and partly intermingled. If attention were to act some- thereby easily addressable maps. Top-down influences among
how directly on those synapses, e.g., through anatomical top- neighboring areas might be refined by the above-discussed
down projections ending in synapses onto those synapses, mechanism that renders feedback from the receiving group
this would require intricate addressing of the synapses more effective at the coherent sending group than at the

226 Neuron 88, October 7, 2015 ª2015 Elsevier Inc.

Neuron

Perspective

A B C D

E F G H

Figure 5. Selective Attention through Selective Interareal Granger-Causal Influences in the Gamma Band
Triplet recording of two sites in V1 (B and F) and one site in V4 (D and H), allowing the analysis of Granger-causal (GC) interareal influences (C and G) (adapted and
modified from Bosman et al., 2012).
(A–D) Two conditions with a single visual stimulus each, showing the stimulus selectivity of the recorded neuronal signals.
(A) Illustration of the two conditions. The stimuli were behaviorally relevant and therefore attended, as indicated by their halo. Red and blue frames are not shown
to the monkey, but are used to label the corresponding spectra in the following panels.
(B) Spectral power changes, relative to pre-stimulus baseline, for the two V1 sites. Each site showed visually induced gamma-band activity exclusively for one of
the two stimuli.
(C) GC influence spectra, showing the feedforward influences of V1 onto V4.
(D) Spectral power change, relative to prestimulus baseline, for the V4 site. The site showed visually induced gamma-band activity that was very similar for the two
stimuli.
(E–H) Same as (A)–(D), but for two conditions with two visual stimuli and selective attention to one of them. (F) In V1, selective attention enhances gamma peak
frequency (see also Figures 6A and 6B). (G) V1 exerts feedforward influence onto V4 almost exclusively through the gamma rhythm induced by the attended
stimulus. (H) V4 responds equally strongly to both conditions. Previous studies have demonstrated that spike rates of single neurons in V4 predominantly
represent the attended stimulus (Moran and Desimone, 1985; Reynolds et al., 1999), as indicated by the stimulus symbols above the panel.

noncoherent sending group, even if it is anatomically directed to with RFs small enough to contain only one of the multiple stimuli,
both (Bastos et al., 2015b; Fries, 2005). because then the competition occurs at the input to the next
This still leaves open the question of how the top-down higher area. For example, two closely spaced stimuli might fall
addressed neuronal groups are brought into selective synchroni- into separate V1 RFs and compete inside the same V4 RF,
zation with the postsynaptic neurons in higher visual areas. A whereas two more separated stimuli might fall into different V4
parsimonious assumption is that attentional top-down influ- RFs and compete inside the same RF in the subsequent area TEO.
ences modify gamma-band synchronization in the lower-area When two stimuli are separated such that a local group of V4
neuronal group in such a way that it has a competitive advantage neurons is activated by only one of them, selective attention to
in entraining the postsynaptic neuronal group (Lee et al., 2013). that stimulus results in enhanced gamma-band synchronization
This might be achieved by modulation in gamma-band synchro- among the V4 neurons (Bichot et al., 2005; Fries et al., 2001),
nization strength or frequency, and I will discuss evidence for which is predictive of shortened reaction times on a trial-by-trial
both in the following two paragraphs. basis (Womelsdorf et al., 2006). This holds at least if gamma-
Selective Attention Modulates Gamma-Band band synchronization is assessed by means of LFP power
Synchronization Strength (Taylor et al., 2005) or the coherence between multiunit activity
The first option is that attentional top-down influence affects (MUA) (Fries et al., 2008). Yet, when MUA, showing an attentional
the strength of gamma-band synchronization in the lower area. enhancement of gamma-band synchronization, is broken
Importantly, in this context, the lower area is simply the area down into single units, a differential effect of attention is revealed

Neuron 88, October 7, 2015 ª2015 Elsevier Inc. 227

 Neuron

Perspective

(Vinck et al., 2013): attention decreases the gamma locking of the gamma-frequency increase. This corresponds well to the finding
less active or less stimulus-activated neurons, whereas it in- that a 50% contrast increase leads to a 4 Hz gamma-frequency
creases the gamma locking of the more active or more stim- increase (Ray and Maunsell, 2010) (the study reported a linear fit
ulus-activated neurons. This increased gamma-band synchroni- between the gamma frequency and log2 [contrast] with a slope of
zation among the most active units likely endows them with a 6.8 Hz).
larger influence onto postsynaptic neurons (Azouz and Gray, Interestingly, gamma peak frequency also systematically in-
2003; Salinas and Sejnowski, 2001). It could thus be a mecha- creases with other stimulus properties that increase stimulus
nism through which attentional top-down mechanisms lend the strength and salience, namely with the elimination of superim-
lower-area neurons representing the attended stimulus a posed noise (Figure 6E) (Jia et al., 2013b), with stimulus motion
competitive advantage in entraining postsynaptic neurons in (Figure 6G) (Friedman-Hill et al., 2000; Gray et al., 1990; Muthu-
higher areas. kumaraswamy and Singh, 2013; Swettenham et al., 2009; van
An interesting question is whether similar effects of attentional Pelt and Fries, 2013), and with a stimulus location closer to the
top-down influences occur in primary visual cortex, V1. When fovea (Figure 6H) (Lima et al., 2010; van Pelt and Fries, 2013).
gamma-band synchronization in V1 is assessed as LFP power Also, stimulus onsets lead to higher gamma peak frequency
or as spike-LFP coherence without differentiating between (Figure 6C), a phenomenon illustrated in many studies even
more versus less active/activated neurons, inconsistent results though not often reported explicitly (Fries et al., 2001; Hoogen-
have been reported: different studies found decreases (Chalk boom et al., 2006). In addition, gamma peak frequency increases
et al., 2010), an absence of a change (Bosman et al., 2012), also with decreasing stimulus size (Figure 6F) (Gieselmann and
and increases (Buffalo et al., 2011; van Kerkoerle et al., 2014). Thiele, 2008; Jia et al., 2013b; Ray and Maunsell, 2011). The rela-
The overall weaker effect of attention on gamma-band synchro- tion of stimulus size to stimulus salience is less clear. Yet, one
nization in V1 as compared to V4 might be due to (1) weaker might speculate that larger stimuli might increasingly be treated
attentional influences, or (2) more pronounced stimulus selec- as nonsalient background. Gamma peak frequency increases
tivity potentially leading to an attentional decrease of gamma are not simply due to higher firing rates. When the salience of
synchronization for a large majority and an increase only for a drifting gratings is reduced by superimposing dynamic noise,
small minority of single neurons. Because in V4 attention leads this is reflected in reduced gamma peak frequency (Figure 6E),
to both gamma-synchronization increases (for strongly active/ but not in firing rate changes (Jia et al., 2013b). Similarly, when
activated neurons) and decreases (for weakly active/activated stimulus orientation is varied around the neuron’s preferred
neurons), it is possible that similar effects occur in V1 and explain orientation, this by definition changes firing rates, but it does
the diverse findings there so far. not change stimulus salience, and correspondingly, it also
Selective Attention and Stimulus Salience Modulate does not systematically affect gamma frequency (Figure 6I)
Gamma-Band Synchronization Frequency (Friedman-Hill et al., 2000; Gray et al., 1990; Jia et al., 2013b).
An additional mechanism that might lend a neuronal group a At any given moment, gamma frequency is likely influenced by
competitive advantage in entraining postsynaptic neurons is an several factors simultaneously. The described effects of stimulus
enhanced gamma-band synchronization frequency. When two salience and attention likely interact with each other and with
simulated oscillators with slightly different intrinsic frequencies further effects. For example, gamma frequency also increases
are reciprocally coupled, they can synchronize fully or partially. with stimulus repetition (Brunet et al., 2014).
During periods of synchronization, the oscillator with the higher I would like to speculate that the competitive advantage of the
intrinsic frequency leads over the one with the lower intrinsic faster V1 gamma is potentiated by a regular reset of gamma
frequency (Cannon et al., 2014; Lowet et al., 2015) and thereby phase by a theta rhythm, as has been found in awake monkey
exerts a relatively stronger influence. Correspondingly, when visual cortex (Bosman et al., 2009). For the first few gamma
two V1 gamma rhythms compete for entraining V4 gamma, it is cycles after the reset, the faster gamma rhythm sends its inputs
conceivable that the faster V1 gamma rhythm exerts a stronger to V4 a few milliseconds before competing slower gamma
influence on V4, simply by means of its higher frequency (Cannon rhythms (Figure 7). The earlier arrival of input from the faster
et al., 2014). gamma allows these inputs to trigger inhibition of the postsyn-
The abovementioned recordings in monkey visual cortex aptic network and thereby to shut out competing inputs medi-
(Bosman et al., 2012) indeed revealed that selective attention ated by slower gamma rhythms. In essence, stimulus salience
leads to a consistent increase in the V1 gamma peak frequency and top-down attention are translated into gamma frequency,
by about 3 Hz (Figures 6A and 6B). Intriguingly, gamma peak and theta-rhythmic gamma phase resets turn V1 gamma-fre-
frequency systematically increases also with increasing stimulus quency differences into V4 input-latency differences, with the
contrast (Figure 6D) (Hadjipapas et al., 2015; Jia et al., 2013b; earliest input representing the most salient and/or attended stim-
Lowet et al., 2015; Ray and Maunsell, 2010; Roberts et al., ulus (Bosman et al., 2012). After a few gamma cycles, once V4 is
2013). Stimulus contrast and selective attention have closely entrained by the winning V1 rhythm, the interareal locking might
related effects on neuronal processing (Reynolds and Chelazzi, sustain the selection for the remainder of the theta cycle, even
2004; Reynolds and Heeger, 2009), with attention increasing when the two gamma rhythms precess against each other and
the effective contrast of an attended stimulus by about 50% thereby produce arbitrary lead-lag relationships. The stability
(Reynolds et al., 2000). If stimulus contrast invoked the same of an established gamma-band entrainment against precession
gamma-frequency-enhancing mechanisms as selective atten- has been suggested by mathematical models (Börgers and
tion, a 50% contrast enhancement should result in a 3 Hz Kopell, 2008).

228 Neuron 88, October 7, 2015 ª2015 Elsevier Inc.

Neuron

Perspective

A B C

D E F

G H I

Figure 6. The Gamma-Band Peak Frequency Increases with Attention and Salience
(A and B) Dots show awake macaque V1 LFP power changes (scaled to peak at a value of one) induced by a grating stimulus, when it was nonattended (blue) or
attended (red), lines show Gaussian fits (adapted and modified from Bosman et al., 2012). (B) shows a detail of (A) at higher resolution.
(C) Human MEG power change over early visual cortex as a function of time after stimulus onset. Note that the gamma peak frequency is higher at response onset
than during the sustained response (adapted and modified from van Pelt et al., 2012).
(D–F) Anesthetized macaque V1 LFP power, during visual stimulation with a grating of varying contrast (D), varying amount of superimposed noise (E), and varying
size (F), as indicated by inset color legends (adapted and modified from Jia et al., 2013b).
(G) Human MEG power change, estimated to emerge from primary visual cortex, during visual stimulation with a grating that is stationary (blue) or moving (red)
(adapted and modified from Swettenham et al., 2009).
(H) Awake macaque V1 LFP power during stimulation with a large grating that activated a recording site with a peripheral RF (blue) and another recording site with
a foveal RF (red) (adapted and modified from Lima et al., 2010).
(I) Anesthetized macaque V1 LFP gamma peak frequency (red line and left y axis) as well as gamma power (blue line and right y axis) as a function of stimulus
orientation relative to the recording site’s preferred orientation (adapted and modified from Jia et al., 2013b).

Stimulus salience, e.g., due to contrast or motion, is often lower-area neuronal group that is activated by the attended
within a narrow range within a given visual object, yet changes stimulus in entraining postsynaptic neurons in higher areas.
at object boundaries (Lowet et al., 2015). Moreover, visual The selective entrainment of the higher area to the part of
attention during natural viewing operates at the level of objects the lower area representing the attended stimulus constitutes
(Nuthmann and Henderson, 2010). Thereby salience- and atten- the implementation of attentional selection. It is important to
tion-dependent gamma frequency changes likely contribute make the distinction between the implementation of atten-
to synchronization of neurons activated by one object and de- tional selection and the control of attentional selection. While
synchronization of neurons driven by different objects (Lowet the control of attentional selection is exerted by frontal and
et al., 2015). Crucially, across the hierarchy of visual areas, parietal areas through top-down influences onto visual areas,
neuronal groups activated by the same visual stimulus naturally the implementation of attentional selection is realized by
see the same stimulus salience and attention and thereby have selective bottom-up influences of lower onto higher visual
the same gamma frequency. Where projections to higher areas areas. Therefore, as much as it makes sense that the imple-
converge, the higher gamma frequency likely succeeds in selec- mentation is through selective interareal gamma-band syn-
tively entraining postsynaptic neurons. chronization, which is predominantly bottom-up, the control
Thus, both the frequency and the strength of gamma-band is expected to be through top-down influences of frontoparie-
synchronization might lend a competitive advantage to the tal areas onto visual areas and/or higher visual areas onto

Neuron 88, October 7, 2015 ª2015 Elsevier Inc. 229

 Neuron

Perspective

spondingly, for those neighboring areas, there are also weak

top-down gamma influences and weak bottom-up beta influ-
ences (Bastos et al., 2015a).
A similar dissociation with higher frequencies mediating feed-
forward and lower frequencies mediating feedback signals has
also been found outside visual cortex. One study investigated
Granger causality between lower and higher auditory cortex of
human patients with implanted electrodes and reported feedfor-
ward signaling in gamma and feedback signaling in delta to beta
bands (Fontolan et al., 2014). Yet, outside visual cortex also
counterexamples have been reported for parietal-auditory and
parietal-somatosensory influences (Brovelli et al., 2004; Roopun
et al., 2010).
Among visual areas, the notion that gamma mediates forward
Figure 7. Theta-Rhythmic Phase Reset Turns Gamma-Frequency and beta mediates backward influences received further support
Differences into Latency Differences
In the lower area, after a reset, the gamma rhythms representing different from a study using electrical stimulation in either V1 or V4 and
stimuli start at the same phase. The gamma rhythm representing the attended LFP recordings in the respective other area (van Kerkoerle
stimulus (green) is faster than the gamma rhythm representing the unattended et al., 2014). When electrical stimulation (five pulses at 200 Hz)
stimulus (orange). This frequency difference translates into a latency differ-
was given to V1, this robustly induced enhanced gamma-band
ence. The input from the attended representation reaches the higher area first,
transmits its representation, and triggers inhibition to shut out the competing LFP power in V4, consistent with a forward influence (Figure 8D);
unattended representation. when the same electrical stimulation was given to V4, then under
visual stimulation with a background stimulus, this induced
lower visual areas (Buffalo et al., 2010; Buschman and Miller, enhanced alpha-band LFP power in V1 (Figure 8E). The fre-
2007). quencies related to top-down influences differ across the studies
between alpha and beta, and the precise reason or relative role
Top-Down Influences Are Mediated of the two neighboring bands is not yet clear.
by Alpha-Beta-Band Rhythms These results predict that top-down beta-band influences
A recent study has investigated the top-down-directed influ- should be enhanced when a cognitive task requires stronger
ences among eight primate visual areas, including parietal and top-down control. A task modulating top-down control is selec-
frontal visual control areas. In the gamma band, bottom-up influ- tive attention, in which both bottom-up stimulation and task dif-
ences were stronger than top-down influences, yet the opposite ficulty are constant, and top-down influences should be specif-
was the case for the beta band, in which top-down influences ically enhanced toward the visual cortex representing the
were stronger than bottom-up influences (Figure 8A) (Bastos attended stimulus. Indeed, top-down beta-band influences
et al., 2015a). This pattern of directed interareal influences among visual areas contralateral to the attended stimulus were
agrees well with another finding, namely that in primate visual found to be larger than among ipsilateral areas (Bastos et al.,
areas V1, V2, and V4, gamma-band synchronization is particu- 2015a). This is consistent with a study reporting that attention
larly strong in superficial cortical layers, whereas alpha/beta- enhances 8–15 Hz coherence between areas V4 and TEO and
band synchronization is particularly strong in deep cortical layers between those areas and the Pulvinar (Saalmann et al., 2012).
(Figure 8C) (Buffalo et al., 2011). Superficial cortical layers are Enhanced top-down beta-band influences might lead to
the primary source of anatomical forward projections, and this enhanced bottom-up gamma-band influences (Bosman et al.,
predominance increases with the number of hierarchical levels 2012; Grothe et al., 2012), and a recent modeling study suggests
that are bridged by the forward projections; deep cortical layers a putative mechanism (Lee et al., 2013). Also during other tasks,
are the primary source of anatomical backward projections, and conditions expected to strengthen feedback influences resulted
this predominance increases with the number of hierarchical in stronger synchronization in relatively lower frequency bands
levels that are bridged by the backward projections (Markov (Arnal et al., 2011; Buschman and Miller, 2007; Kornblith et al.,
et al., 2014). Thus, the degree to which an anatomical projection 2015; von Stein et al., 2000).
originates from superficial layers, measured by the supragranu-
lar labeled neuron proportion after retrograde tracing (SLN), Theta-Rhythmic Gamma-Band Synchronization
quantifies the degree to which a projection is of a feedforward Implements Attentional Sampling
type (Barone et al., 2000). Consistent with the laminar differences The experimental evidence presented and the considerations
in gamma- versus beta-band synchronization, the anatomical discussed so far suggest that top-down attentional influences
SLN metric correlates with the asymmetry in directed influences are mediated by beta-band synchronization, that the selective
(Figure 8B) (Bastos et al., 2015a). These data suggest that supra- communication of the attended stimulus is implemented by
granular layers convey their signals through the gamma rhythm, gamma-band synchronization, and that gamma is rhythmically
whereas infragranular layers use the beta rhythm. Note that weak reset by a 4 Hz theta rhythm. The theta-rhythmic resetting entails
top-down projections can originate in supragranular layers and a modulation of gamma-synchronization strength, consistent
weak bottom-up projections in infragranular layers, for areas with several studies reporting gamma-strength modulation with
close to each other in the hierarchy (Markov et al., 2014). Corre- theta phase (Figure 9A) (Bosman et al., 2009, 2012; Bragin

230 Neuron 88, October 7, 2015 ª2015 Elsevier Inc.

Neuron

Perspective

A et al., 1995; Canolty et al., 2006; Colgin et al., 2009; Schomburg

et al., 2014; Voloh et al., 2015). If we identify a gamma-synchro-
nized network with the attentional selection of the respective
stimulus, then the theta-rhythmic gamma resetting corresponds
to the termination of this attentional selection and a potential shift
of attention to another stimulus. Attentional shifts are expressed
overtly during natural viewing as gaze shifts, i.e., saccades.
Strikingly, saccades during natural viewing show intersaccadic
interval distributions with a strong Gaussian component at
145 ms (Figures 9B and 9C), consistent with a 7 Hz theta rhythm
(Otero-Millan et al., 2008).
B Together, these data suggest that the theta rhythm constitutes
a visual (and maybe a general) exploration routine (Fries, 2009).
When the behavioral context allows the eyes to explore the
visual environment freely, they saccade at a theta rhythm.
When the context requires the eyes to fixate for prolonged
durations, the cortical theta phase modulates the strength of
gamma-band synchronization (Bosman et al., 2009), which we
identified as implementation of attentional selection. Thus, it is
conceivable that irrespective of saccades as overt expressions
of attention, selective attention samples visual input at a theta
C rhythm.
A recent study lends direct support to this (Landau and Fries,
2012). Human subjects were required to fixate and report a
contrast decrement occurring at a random time per trial in one
of two gratings positioned in the right and left visual hemifield.
A task-irrelevant flash occurred around one of the gratings at a
random time relative to the to-be-detected contrast decrement
and shifted attention to the respective grating. This attentional
reset to one side was particularly successful when the flash
occurred in the right visual field. After right visual field flashes,
D 4 Hz theta rhythms were directly visible in the time courses of
behavioral performance at both stimulus locations, and the two
rhythms were in antiphase (Figures 9D and 9E). Another study
used a similar approach to investigate detection performance
at uncued locations, at which changes occurred with 25% prob-
ability, and which were either on the same or a different object
as the cued location, at which changes occurred with 75%
probability (Fiebelkorn et al., 2013). This revealed attentional
sampling of the uncued locations at 4 Hz and in antiphase. Yet

E purple and orange backgrounds, respectively. The influence in the DP-to-V1

direction is through an anatomical feedback-type projection and pre-
dominates in the beta band, indicated by green background.
(B) The Spearman-rank correlation, across area pairs, between an anatomical
metric of the feedforward/feedback character of an interareal projection (SLN)
and an electrophysiological metric of the asymmetry in Granger-causal in-
fluences (DAI). A positive (negative) correlation value indicates that Granger-
causal influences in the respective frequency are stronger in the anatomically
defined feedforward (feedback) direction. (A) and (B) are adapted and modified
from Bastos et al. (2015a).
(C) Spike-LFP coherence from awake macaque area V2, for recordings
from deep (blue) and superficial (red) layers. Spike-LFP coherence shows an
alpha-beta band peak for deep layers and both a theta and a gamma peak for
superficial layers (adapted and modified from Buffalo et al., 2011).
(D) Awake macaque V4 LFP power during visual stimulation with a background
stimulus (black) and additional electrical stimulation in V1 (five pulses at
Figure 8. Feedforward Predominates in Theta and Gamma Bands, 200 Hz), which leads to power enhancement in the gamma band (red).
Feedback in the Alpha-Beta Band (E) Awake macaque V1 LFP power during visual stimulation with a background
(A) Granger-causal influences between awake macaque areas V1 and DP. The stimulus (black) and additional electrical stimulation in V4 (five pulses at
influence in the V1-to-DP direction is through an anatomical feedforward-type 200 Hz), which leads to power enhancement in the alpha-beta band (blue). (D)
projection and predominates in the theta and gamma bands, indicated by and (E) are adapted and modified from van Kerkoerle et al. (2014).

Neuron 88, October 7, 2015 ª2015 Elsevier Inc. 231

 Neuron

Perspective

A B C Figure 9. A Theta Rhythm Is Visible in Visual

Gamma, Saccades, and Attentional
Sampling
(A) V1–V4 coherence as a function of frequency
and of time in the 4 Hz theta cycle (adapted and
modified from Bosman et al., 2012).
(B) Histograms of intersaccadic intervals (ISIs) for
microsaccades and regular saccades during free
viewing of natural scenes.
(C) The histograms of (B) have been fitted with
ex-Gaussian functions. The resulting parameter
estimates are shown for microsaccades and
saccades observed during several viewing con-
ditions as indicated. Irrespective of condition, the
Gaussian component’s mean was around 145 ms,
corresponding to 7 Hz. (B) and (C) are adapted and
modified from Otero-Millan et al. (2008).
(D) Detection accuracy for equally probable
contrast decrements on two bilateral stimuli, after
D E an irrelevant flash at time zero.
(E) Spectral analysis of the time-resolved detec-
tion accuracy from (D). The amplitude spectra
reveal peaks close to 4 Hz. The 4 Hz phase relation
of detection accuracy contralateral versus ipsilat-
eral to the flash is shown in the inset, has a mean of
222 , and is not significantly different from 180 . (D
and E) are adapted and modified from Landau and
Fries (2012).

Conclusions
Based on the synopsis of the presented
evidence and considerations, I would
like to suggest the following scenario:
another very recent study demonstrated that the 4 Hz sampling Local cortical neuronal groups synchronize by default in the
of two simultaneously attended stimuli does not only occur after alpha band. During alpha-band synchronization, network excita-
a reset event, but that it is present continuously during distrib- tion fluctuates at 100 ms cycles, but is tracked by network inhi-
uted attention (Landau et al., 2015). Subjects monitored two bition within 3 ms. This curtails effective communication and ren-
equally relevant stimuli for an unpredictable small change, and ders the respective activity invisible to other neurons. It allows
the gamma-band activities induced by the two stimuli were holding ‘‘on-stock’’ local neuronal representation, which can
subtracted from each other to reveal moment-by-moment atten- be accessed flexibly. It might be an important contribution to
tional biases to one or the other stimulus. The 4 Hz phase of this making optimal use of the brain’s massively parallel processing
gamma difference preceding the stimulus change predicted architecture. Attention samples from this internal store at a theta
whether subjects were better or worse in detecting the change. rhythm. Attentional top-down influences are mediated by alpha-
This suggests that the gamma-band activities induced by the beta-band synchronization. It is possible that top-down influ-
two stimuli are enhanced at 4 Hz in alternation, whenever the ences in alpha and beta bands have differential roles. Top-
respective stimulus is attentionally sampled. down alpha influences might convey influences that reinforce
These results can be parsimoniously explained by an 8 Hz local alpha, e.g., for irrelevant background regions. Top-down
attentional sampling process. If this process samples two stim- beta influences wake up the local circuit and modulate its
uli, each stimulus is sampled four times per second, i.e., at a gamma-band synchronization by strengthening it among the
4 Hz rhythm. This would predict that a single object would be most stimulus-driven neurons, and by enhancing its frequency.
sampled at around 8 Hz and, e.g., three stimuli at around 8/3 = Visual scenes induce many local gamma rhythms with varying
2.7 Hz. Indeed, detection performance of a single location strength and frequency, reflecting the bottom-up stimulus
fluctuates and can be partly predicted by the prestimulus phase salience and stimulus history. The resulting gamma landscape
of the ongoing 7.1 Hz component of frontal EEG (Busch et al., in, e.g., V1 thus reflects stimulus properties, experience, and
2009). An elegant psychophysical study investigated human top-down influences. At a given time point, one out of these co-
tracking performance on one, two, and three moving targets existing gamma rhythms succeeds in entraining postsynaptic
while varying temporal frequency of the tracked stimuli. This neuronal groups. This gamma entrainment allows to transmit a
revealed that temporal frequency limits fell from 7 Hz with one stimulus representation and to selfishly shut out competing stim-
target to 4 Hz with two targets and 2.6 Hz with three targets uli’s representations. The entrainment establishes a cycle-to-cy-
(Holcombe and Chen, 2013). These results are consistent with cle memory of the active link that maintains until it is terminated
a 7–8 Hz sampling process that is divided over one, two, or three at the end of a theta cycle. The presynaptic gamma rhythm al-
stimuli. lows network excitation to escape its ever-chasing network

232 Neuron 88, October 7, 2015 ª2015 Elsevier Inc.

Neuron

Perspective

inhibition. Inhibition is temporally focused, which allows excita- Bragin, A., Jandó, G., Nádasdy, Z., Hetke, J., Wise, K., and Buzsáki, G. (1995).
Gamma (40-100 Hz) oscillation in the hippocampus of the behaving rat.
tion in between inhibition, and which synchronizes excitatory J. Neurosci. 15, 47–60.
output to optimally drive postsynaptic neurons. The rhythmically
synchronous communication establishes a pulsatile computa- Brovelli, A., Ding, M., Ledberg, A., Chen, Y., Nakamura, R., and Bressler, S.L.
(2004). Beta oscillations in a large-scale sensorimotor cortical network: direc-
tion. During each pulse, a spike-rate-based stimulus representa- tional influences revealed by Granger causality. Proc. Natl. Acad. Sci. USA
tion passes through the respective synapses to a postsynaptic 101, 9849–9854.
neuron. The temporal focusing in the pulse increases the
Brunet, N.M., Bosman, C.A., Vinck, M., Roberts, M., Oostenveld, R., Desi-
precision with which postsynaptic currents combine. Neurons mone, R., De Weerd, P., and Fries, P. (2014). Stimulus repetition modulates
driven by preferred stimuli spike slightly earlier in the pulse, gamma-band synchronization in primate visual cortex. Proc. Natl. Acad. Sci.
USA 111, 3626–3631.
with higher postsynaptic impact. Thus, several rhythms and
their interplay render neuronal communication effective, precise, Brunet, N., Bosman, C.A., Roberts, M., Oostenveld, R., Womelsdorf, T., De
and selective. Weerd, P., and Fries, P. (2015). Visual cortical gamma-band activity during
free viewing of natural images. Cereb. Cortex 25, 918–926.

Buffalo, E.A., Fries, P., Landman, R., Liang, H., and Desimone, R. (2010).
ACKNOWLEDGMENTS
A backward progression of attentional effects in the ventral stream. Proc.
Natl. Acad. Sci. USA 107, 361–365.
I thank the members of my lab for contributing to the development of CTC
and its experimental tests. I thank Nancy Kopell, Ayelet Landau, Christopher Buffalo, E.A., Fries, P., Landman, R., Buschman, T.J., and Desimone, R.
Lewis, Eric Lowet, Eric Maris, Alina Peter, Michael Schmid, Wolf Singer, (2011). Laminar differences in gamma and alpha coherence in the ventral
Julien Vezoli, Martin Vinck, and Thilo Womelsdorf for helpful comments on stream. Proc. Natl. Acad. Sci. USA 108, 11262–11267.
the manuscript. This work was supported by a European Young Investigator
Award, HFSP (RGP0070/2003), Volkswagen Foundation (I/79876), European Busch, N.A., Dubois, J., and VanRullen, R. (2009). The phase of ongoing EEG
Union (HEALTH-F2-2008-200728 and HBP-604102), NIH (1U54MH091657), oscillations predicts visual perception. J. Neurosci. 29, 7869–7876.
DFG (FOR 1847), and LOEWE (NeFF).
Buschman, T.J., and Miller, E.K. (2007). Top-down versus bottom-up control
of attention in the prefrontal and posterior parietal cortices. Science 315,
1860–1862.
REFERENCES
Buzsáki, G., and Wang, X.J. (2012). Mechanisms of gamma oscillations. Annu.
Ainsworth, M., Lee, S., Cunningham, M.O., Traub, R.D., Kopell, N.J., and Rev. Neurosci. 35, 203–225.
Whittington, M.A. (2012). Rates and rhythms: a synergistic view of frequency
and temporal coding in neuronal networks. Neuron 75, 572–583. Cannon, J., McCarthy, M.M., Lee, S., Lee, J., Börgers, C., Whittington, M.A.,
and Kopell, N. (2014). Neurosystems: brain rhythms and cognitive processing.
Akam, T.E., and Kullmann, D.M. (2012). Efficient ‘‘communication through Eur. J. Neurosci. 39, 705–719.
coherence’’ requires oscillations structured to minimize interference between
signals. PLoS Comput. Biol. 8, e1002760. Canolty, R.T., Edwards, E., Dalal, S.S., Soltani, M., Nagarajan, S.S., Kirsch, H.E.,
Berger, M.S., Barbaro, N.M., and Knight, R.T. (2006). High gamma power is
Arnal, L.H., Wyart, V., and Giraud, A.L. (2011). Transitions in neural oscillations phase-locked to theta oscillations in human neocortex. Science 313, 1626–1628.
reflect prediction errors generated in audiovisual speech. Nat. Neurosci. 14,
797–801. Cardin, J.A., Carlén, M., Meletis, K., Knoblich, U., Zhang, F., Deisseroth, K.,
Tsai, L.H., and Moore, C.I. (2009). Driving fast-spiking cells induces gamma
Atallah, B.V., and Scanziani, M. (2009). Instantaneous modulation of gamma rhythm and controls sensory responses. Nature 459, 663–667.
oscillation frequency by balancing excitation with inhibition. Neuron 62,
566–577. Chalk, M., Herrero, J.L., Gieselmann, M.A., Delicato, L.S., Gotthardt, S., and
Thiele, A. (2010). Attention reduces stimulus-driven gamma frequency oscilla-
Azouz, R., and Gray, C.M. (2003). Adaptive coincidence detection and dy- tions and spike field coherence in V1. Neuron 66, 114–125.
namic gain control in visual cortical neurons in vivo. Neuron 37, 513–523.
Colgin, L.L., Denninger, T., Fyhn, M., Hafting, T., Bonnevie, T., Jensen, O.,
Barone, P., Batardiere, A., Knoblauch, K., and Kennedy, H. (2000). Laminar Moser, M.B., and Moser, E.I. (2009). Frequency of gamma oscillations routes
distribution of neurons in extrastriate areas projecting to visual areas V1 and flow of information in the hippocampus. Nature 462, 353–357.
V4 correlates with the hierarchical rank and indicates the operation of a dis-
tance rule. J. Neurosci. 20, 3263–3281. Csicsvari, J., Jamieson, B., Wise, K.D., and Buzsáki, G. (2003). Mechanisms of
gamma oscillations in the hippocampus of the behaving rat. Neuron 37,
Bastos, A.M., Vezoli, J., Bosman, C.A., Schoffelen, J.M., Oostenveld, R., Dow- 311–322.
dall, J.R., De Weerd, P., Kennedy, H., and Fries, P. (2015a). Visual areas exert
feedforward and feedback influences through distinct frequency channels. Felleman, D.J., and Van Essen, D.C. (1991). Distributed hierarchical process-
Neuron 85, 390–401. ing in the primate cerebral cortex. Cereb. Cortex 1, 1–47.

Bastos, A.M., Vezoli, J., and Fries, P. (2015b). Communication through coher- Fiebelkorn, I.C., Saalmann, Y.B., and Kastner, S. (2013). Rhythmic sampling
ence with inter-areal delays. Curr. Opin. Neurobiol. 31, 173–180. within and between objects despite sustained attention at a cued location.
Curr. Biol. 23, 2553–2558.
Bichot, N.P., Rossi, A.F., and Desimone, R. (2005). Parallel and serial neural
mechanisms for visual search in macaque area V4. Science 308, 529–534. Fontolan, L., Morillon, B., Liegeois-Chauvel, C., and Giraud, A.L. (2014). The
contribution of frequency-specific activity to hierarchical information process-
Börgers, C., and Kopell, N.J. (2008). Gamma oscillations and stimulus selec- ing in the human auditory cortex. Nat. Commun. 5, 4694.
tion. Neural Comput. 20, 383–414.
Friedman-Hill, S., Maldonado, P.E., and Gray, C.M. (2000). Dynamics of striate
Bosman, C.A., Womelsdorf, T., Desimone, R., and Fries, P. (2009). A micro- cortical activity in the alert macaque: I. Incidence and stimulus-dependence of
saccadic rhythm modulates gamma-band synchronization and behavior. gamma-band neuronal oscillations. Cereb. Cortex 10, 1105–1116.
J. Neurosci. 29, 9471–9480.
Fries, P. (2005). A mechanism for cognitive dynamics: neuronal communica-
Bosman, C.A., Schoffelen, J.M., Brunet, N., Oostenveld, R., Bastos, A.M., tion through neuronal coherence. Trends Cogn. Sci. 9, 474–480.
Womelsdorf, T., Rubehn, B., Stieglitz, T., De Weerd, P., and Fries, P. (2012).
Attentional stimulus selection through selective synchronization between Fries, P. (2009). Neuronal gamma-band synchronization as a fundamental pro-
monkey visual areas. Neuron 75, 875–888. cess in cortical computation. Annu. Rev. Neurosci. 32, 209–224.

Neuron 88, October 7, 2015 ª2015 Elsevier Inc. 233

 Neuron

Perspective
Fries, P., Reynolds, J.H., Rorie, A.E., and Desimone, R. (2001). Modulation of Lee, J.H., Whittington, M.A., and Kopell, N.J. (2013). Top-down beta rhythms
oscillatory neuronal synchronization by selective visual attention. Science 291, support selective attention via interlaminar interaction: a model. PLoS Com-
1560–1563. put. Biol. 9, e1003164.

, D., and Singer, W. (2007). The gamma cycle. Trends Neuro-

Fries, P., Nikolic Lima, B., Singer, W., Chen, N.H., and Neuenschwander, S. (2010). Synchroni-
sci. 30, 309–316. zation dynamics in response to plaid stimuli in monkey V1. Cereb. Cortex 20,
1556–1573.
Fries, P., Womelsdorf, T., Oostenveld, R., and Desimone, R. (2008). The effects
of visual stimulation and selective visual attention on rhythmic neuronal syn- Livingstone, M.S. (1996). Oscillatory firing and interneuronal correlations in
chronization in macaque area V4. J. Neurosci. 28, 4823–4835. squirrel monkey striate cortex. J. Neurophysiol. 75, 2467–2485.

Gielen, S., Krupa, M., and Zeitler, M. (2010). Gamma oscillations as a mecha- Lowet, E., Roberts, M., Hadjipapas, A., Peter, A., van der Eerden, J., and De
nism for selective information transmission. Biol. Cybern. 103, 151–165. Weerd, P. (2015). Input-dependent frequency modulation of cortical gamma
oscillations shapes spatial synchronization and enables phase coding. PLoS
Gieselmann, M.A., and Thiele, A. (2008). Comparison of spatial integration and Comput. Biol. 11, e1004072.
surround suppression characteristics in spiking activity and the local field
potential in macaque V1. Eur. J. Neurosci. 28, 447–459. Luck, S.J., Chelazzi, L., Hillyard, S.A., and Desimone, R. (1997). Neural mech-
anisms of spatial selective attention in areas V1, V2, and V4 of macaque visual
Gray, C.M., König, P., Engel, A.K., and Singer, W. (1989). Oscillatory re- cortex. J. Neurophysiol. 77, 24–42.
sponses in cat visual cortex exhibit inter-columnar synchronization which
reflects global stimulus properties. Nature 338, 334–337. Markov, N.T., Vezoli, J., Chameau, P., Falchier, A., Quilodran, R., Huissoud, C.,
Lamy, C., Misery, P., Giroud, P., Ullman, S., et al. (2014). Anatomy of hierarchy:
Gray, C.M., Engel, A.K., König, P., and Singer, W. (1990). Stimulus-dependent feedforward and feedback pathways in macaque visual cortex. J. Comp. Neu-
neuronal oscillations in cat visual cortex: receptive field properties and feature rol. 522, 225–259.
dependence. Eur. J. Neurosci. 2, 607–619.
Moran, J., and Desimone, R. (1985). Selective attention gates visual process-
Gregoriou, G.G., Gotts, S.J., Zhou, H., and Desimone, R. (2009). High-fre- ing in the extrastriate cortex. Science 229, 782–784.
quency, long-range coupling between prefrontal and visual cortex during
attention. Science 324, 1207–1210. Muthukumaraswamy, S.D., and Singh, K.D. (2013). Visual gamma oscillations:
the effects of stimulus type, visual field coverage and stimulus motion on MEG
Grothe, I., Neitzel, S.D., Mandon, S., and Kreiter, A.K. (2012). Switching
and EEG recordings. Neuroimage 69, 223–230.
neuronal inputs by differential modulations of gamma-band phase-coherence.
J. Neurosci. 32, 16172–16180. Nuthmann, A., and Henderson, J.M. (2010). Object-based attentional selection
in scene viewing. J. Vis. 10, 20.
Hadjipapas, A., Lowet, E., Roberts, M.J., Peter, A., and De Weerd, P. (2015).
Parametric variation of gamma frequency and power with luminance contrast: Otero-Millan, J., Troncoso, X.G., Macknik, S.L., Serrano-Pedraza, I., and
a comparative study of human MEG and monkey LFP and spike responses. Martinez-Conde, S. (2008). Saccades and microsaccades during visual fixa-
Neuroimage 112, 327–340. tion, exploration, and search: foundations for a common saccadic generator.
J. Vis. 8, 21.21-18.
Haider, B., Duque, A., Hasenstaub, A.R., and McCormick, D.A. (2006).
Neocortical network activity in vivo is generated through a dynamic balance
Ray, S., and Maunsell, J.H. (2010). Differences in gamma frequencies across
of excitation and inhibition. J. Neurosci. 26, 4535–4545.
visual cortex restrict their possible use in computation. Neuron 67, 885–896.
Hasenstaub, A., Shu, Y., Haider, B., Kraushaar, U., Duque, A., and
Ray, S., and Maunsell, J.H. (2011). Different origins of gamma rhythm and high-
McCormick, D.A. (2005). Inhibitory postsynaptic potentials carry synchronized
gamma activity in macaque visual cortex. PLoS Biol. 9, e1000610.
frequency information in active cortical networks. Neuron 47, 423–435.

, D.
Havenith, M.N., Yu, S., Biederlack, J., Chen, N.H., Singer, W., and Nikolic Renart, A., de la Rocha, J., Bartho, P., Hollender, L., Parga, N., Reyes, A., and
(2011). Synchrony makes neurons fire in sequence, and stimulus properties Harris, K.D. (2010). The asynchronous state in cortical circuits. Science 327,
determine who is ahead. J. Neurosci. 31, 8570–8584. 587–590.

Holcombe, A.O., and Chen, W.Y. (2013). Splitting attention reduces temporal Reynolds, J.H., and Chelazzi, L. (2004). Attentional modulation of visual pro-
resolution from 7 Hz for tracking one object to <3 Hz when tracking three. cessing. Annu. Rev. Neurosci. 27, 611–647.
J. Vis. 13, 12.
Reynolds, J.H., and Heeger, D.J. (2009). The normalization model of attention.
Hoogenboom, N., Schoffelen, J.M., Oostenveld, R., Parkes, L.M., and Fries, P. Neuron 61, 168–185.
(2006). Localizing human visual gamma-band activity in frequency, time and
space. Neuroimage 29, 764–773. Reynolds, J.H., Chelazzi, L., and Desimone, R. (1999). Competitive mecha-
nisms subserve attention in macaque areas V2 and V4. J. Neurosci. 19,
Ito, M., Tamura, H., Fujita, I., and Tanaka, K. (1995). Size and position invariance 1736–1753.
of neuronal responses in monkey inferotemporal cortex. J. Neurophysiol. 73,
218–226. Reynolds, J.H., Pasternak, T., and Desimone, R. (2000). Attention increases
sensitivity of V4 neurons. Neuron 26, 703–714.
Jensen, O., and Mazaheri, A. (2010). Shaping functional architecture by oscil-
latory alpha activity: gating by inhibition. Front. Hum. Neurosci. 4, 186. Roberts, M.J., Lowet, E., Brunet, N.M., Ter Wal, M., Tiesinga, P., Fries, P., and
De Weerd, P. (2013). Robust gamma coherence between macaque V1 and V2
Jia, X., Tanabe, S., and Kohn, A. (2013a). g and the coordination of spiking by dynamic frequency matching. Neuron 78, 523–536.
activity in early visual cortex. Neuron 77, 762–774.
Roopun, A.K., Lebeau, F.E., Rammell, J., Cunningham, M.O., Traub, R.D., and
Jia, X., Xing, D., and Kohn, A. (2013b). No consistent relationship between Whittington, M.A. (2010). Cholinergic neuromodulation controls directed
gamma power and peak frequency in macaque primary visual cortex. temporal communication in neocortex in vitro. Front. Neural Circuits 4, 8.
J. Neurosci. 33, 17–25.
Saalmann, Y.B., Pinsk, M.A., Wang, L., Li, X., and Kastner, S. (2012). The
Kornblith, S., Buschman, T.J., and Miller, E.K. (2015). Stimulus load and oscil- pulvinar regulates information transmission between cortical areas based on
latory activity in higher cortex. Cereb. Cortex, bhv182. attention demands. Science 337, 753–756.

Landau, A.N., and Fries, P. (2012). Attention samples stimuli rhythmically. Curr. Salinas, E., and Sejnowski, T.J. (2001). Correlated neuronal activity and the
Biol. 22, 1000–1004. flow of neural information. Nat. Rev. Neurosci. 2, 539–550.

Landau, A.N., Schreyer, H.M., van Pelt, S., and Fries, P. (2015). Distributed Salkoff, D.B., Zagha, E., Yüzgeç, Ö., and McCormick, D.A. (2015). Synaptic
attention is implemented through theta-rhythmic gamma modulation. Curr. Mechanisms of Tight Spike Synchrony at Gamma Frequency in Cerebral
Biol. 25, 2332–2337. Cortex. J. Neurosci. 35, 10236–10251.

234 Neuron 88, October 7, 2015 ª2015 Elsevier Inc.

Neuron

Perspective
Schoffelen, J.M., Oostenveld, R., and Fries, P. (2005). Neuronal coherence as van Pelt, S., and Fries, P. (2013). Visual stimulus eccentricity affects human
a mechanism of effective corticospinal interaction. Science 308, 111–113. gamma peak frequency. Neuroimage 78, 439–447.

Schoffelen, J.M., Poort, J., Oostenveld, R., and Fries, P. (2011). Selective van Pelt, S., Boomsma, D.I., and Fries, P. (2012). Magnetoencephalography in
movement preparation is subserved by selective increases in corticomuscular twins reveals a strong genetic determination of the peak frequency of visually
gamma-band coherence. J. Neurosci. 31, 6750–6758. induced g-band synchronization. J. Neurosci. 32, 3388–3392.
Schomburg, E.W., Fernández-Ruiz, A., Mizuseki, K., Berényi, A., Anastassiou, Vinck, M., Lima, B., Womelsdorf, T., Oostenveld, R., Singer, W., Neuensch-
C.A., Koch, C., and Buzsáki, G. (2014). Theta phase segregation of input- wander, S., and Fries, P. (2010). Gamma-phase shifting in awake monkey
specific gamma patterns in entorhinal-hippocampal networks. Neuron 84, visual cortex. J. Neurosci. 30, 1250–1257.
470–485.
Vinck, M., Womelsdorf, T., Buffalo, E.A., Desimone, R., and Fries, P. (2013).
Shu, Y., Hasenstaub, A., and McCormick, D.A. (2003). Turning on and off
Attentional modulation of cell-class-specific gamma-band synchronization in
recurrent balanced cortical activity. Nature 423, 288–293.
awake monkey area v4. Neuron 80, 1077–1089.
Siegle, J.H., Pritchett, D.L., and Moore, C.I. (2014). Gamma-range synchroni-
zation of fast-spiking interneurons can enhance detection of tactile stimuli. Voloh, B., Valiante, T.A., Everling, S., and Womelsdorf, T. (2015). Theta-
Nat. Neurosci. 17, 1371–1379. gamma coordination between anterior cingulate and prefrontal cortex indexes
correct attention shifts. Proc. Natl. Acad. Sci. USA 112, 8457–8462.
Swettenham, J.B., Muthukumaraswamy, S.D., and Singh, K.D. (2009). Spec-
tral properties of induced and evoked gamma oscillations in human early vi- von Stein, A., Chiang, C., and König, P. (2000). Top-down processing medi-
sual cortex to moving and stationary stimuli. J. Neurophysiol. 102, 1241– ated by interareal synchronization. Proc. Natl. Acad. Sci. USA 97, 14748–
1253. 14753.

Tan, A.Y., Chen, Y., Scholl, B., Seidemann, E., and Priebe, N.J. (2014). Sensory Womelsdorf, T., Fries, P., Mitra, P.P., and Desimone, R. (2006). Gamma-band
stimulation shifts visual cortex from synchronous to asynchronous states. synchronization in visual cortex predicts speed of change detection. Nature
Nature 509, 226–229. 439, 733–736.

Taylor, K., Mandon, S., Freiwald, W.A., and Kreiter, A.K. (2005). Coherent Womelsdorf, T., Schoffelen, J.M., Oostenveld, R., Singer, W., Desimone, R.,
oscillatory activity in monkey area v4 predicts successful allocation of atten- Engel, A.K., and Fries, P. (2007). Modulation of neuronal interactions through
tion. Cereb. Cortex 15, 1424–1437. neuronal synchronization. Science 316, 1609–1612.
Tiesinga, P.H., Fellous, J.M., Salinas, E., José, J.V., and Sejnowski, T.J. (2004). Womelsdorf, T., Lima, B., Vinck, M., Oostenveld, R., Singer, W., Neuensch-
Inhibitory synchrony as a mechanism for attentional gain modulation. wander, S., and Fries, P. (2012). Orientation selectivity and noise correlation
J. Physiol. Paris 98, 296–314. in awake monkey area V1 are modulated by the gamma cycle. Proc. Natl.
Acad. Sci. USA 109, 4302–4307.
van Dijk, H., Schoffelen, J.M., Oostenveld, R., and Jensen, O. (2008).
Prestimulus oscillatory activity in the alpha band predicts visual discrimination
Worden, M.S., Foxe, J.J., Wang, N., and Simpson, G.V. (2000). Anticipatory
ability. J. Neurosci. 28, 1816–1823.
biasing of visuospatial attention indexed by retinotopically specific alpha-
van Ede, F., de Lange, F., Jensen, O., and Maris, E. (2011). Orienting attention band electroencephalography increases over occipital cortex. J. Neurosci.
to an upcoming tactile event involves a spatially and temporally specific mod- 20, RC63.
ulation of sensorimotor alpha- and beta-band oscillations. J. Neurosci. 31,
2016–2024. Zandvakili, A., and Kohn, A. (2015). Coordinated neuronal activity enhances
corticocortical communication. Neuron 87, 827–839.
van Elswijk, G., Maij, F., Schoffelen, J.M., Overeem, S., Stegeman, D.F., and
Fries, P. (2010). Corticospinal beta-band synchronization entails rhythmic Zhang, Y., Meyers, E.M., Bichot, N.P., Serre, T., Poggio, T.A., and Desimone,
gain modulation. J. Neurosci. 30, 4481–4488. R. (2011). Object decoding with attention in inferior temporal cortex. Proc. Natl.
Acad. Sci. USA 108, 8850–8855.
van Kerkoerle, T., Self, M.W., Dagnino, B., Gariel-Mathis, M.A., Poort, J., van
der Togt, C., and Roelfsema, P.R. (2014). Alpha and gamma oscillations Zumer, J.M., Scheeringa, R., Schoffelen, J.M., Norris, D.G., and Jensen, O.
characterize feedback and feedforward processing in monkey visual cortex. (2014). Occipital alpha activity during stimulus processing gates the informa-
Proc. Natl. Acad. Sci. USA 111, 14332–14341. tion flow to object-selective cortex. PLoS Biol. 12, e1001965.

Neuron 88, October 7, 2015 ª2015 Elsevier Inc. 235