12.01.2013 Views

Aligning the Brain in a Rhythmic World

Aligning the Brain in a Rhythmic World

Aligning the Brain in a Rhythmic World

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

© 2009 Schroeder<br />

<strong>Align<strong>in</strong>g</strong> <strong>the</strong> <strong>Bra<strong>in</strong></strong> <strong>in</strong> a <strong>Rhythmic</strong> <strong>World</strong><br />

Charles e. Schroeder, PhD, 1,2 Peter Lakatos, PhD, 1<br />

Chi-M<strong>in</strong>g Chen, PhD, 2 Thomas Radman, PhD, 1<br />

and Anne-Marie Barczak 1,3<br />

1 Cognitive Neuroscience and Schizophrenia Program<br />

Nathan S. Kl<strong>in</strong>e Institute for Psychiatric Research<br />

Orangeburg, New York<br />

2 Department of Psychiatry, Columbia University College of Physicians and Surgeons<br />

New York, New York<br />

3 Department of Physiology, New York University School of Medic<strong>in</strong>e<br />

New York, New York


Introduction: Internal and<br />

External Rhythms<br />

Rhythm <strong>in</strong> <strong>the</strong> bra<strong>in</strong><br />

Except under extreme conditions, such as anes<strong>the</strong>sia,<br />

electrical record<strong>in</strong>gs from <strong>the</strong> bra<strong>in</strong> <strong>in</strong>variably reveal<br />

robust oscillatory activity (fluctuations of voltage). In<br />

pursuit of <strong>the</strong> idea that <strong>the</strong>se oscillations are actually<br />

relevant to adaptive bra<strong>in</strong> function<strong>in</strong>g, <strong>in</strong>creas<strong>in</strong>gly<br />

specific relationships have been proposed between<br />

neuronal oscillations and numerous cognitive<br />

functions. These relationships <strong>in</strong>clude: (1) <strong>the</strong>ta<br />

oscillations and encod<strong>in</strong>g of spatial <strong>in</strong>formation <strong>in</strong><br />

<strong>the</strong> hippocampus (Maurer and McNaughten, 2007);<br />

(2) <strong>the</strong>ta oscillations and <strong>the</strong> formation of mnemonic<br />

neuronal representations (Jensen et al., 2007);<br />

(3) alpha oscillations and “<strong>in</strong>ternally directed”<br />

cognition (Palva and Palva, 2008); and (4) gamma<br />

oscillations and both attention/sensory selection<br />

(Fries et al., 2002) and feature b<strong>in</strong>d<strong>in</strong>g (S<strong>in</strong>ger and<br />

Gray, 1995). See <strong>the</strong> short course chapter Rhythms<br />

<strong>in</strong> Cognitive Process<strong>in</strong>g by C. Tallon-Baudry, for<br />

<strong>in</strong>-depth discussion of this subject. Thus, consensus is<br />

build<strong>in</strong>g that neuronal oscillations are <strong>in</strong>strumental<br />

ra<strong>the</strong>r than <strong>in</strong>cidental to bra<strong>in</strong> function and<br />

dysfunction (Buzsaki and Draguhn, 2004; Uhlhaas<br />

and S<strong>in</strong>ger, 2006; Buzsaki, 2007; Borgers and Kopell,<br />

2008; Schroeder and Lakatos, 2009a).<br />

Rhythm <strong>in</strong> <strong>the</strong> sensorium<br />

When consider<strong>in</strong>g <strong>the</strong> functional significance of<br />

neuronal oscillations, it is worth keep<strong>in</strong>g <strong>in</strong> m<strong>in</strong>d<br />

just how rhythmic our sensory world is. Under<br />

natural conditions, for example, stimulat<strong>in</strong>g <strong>the</strong><br />

hand typically <strong>in</strong>volves mov<strong>in</strong>g <strong>the</strong> f<strong>in</strong>gers across<br />

surfaces or manipulat<strong>in</strong>g objects. Because <strong>the</strong> motor<br />

rout<strong>in</strong>es that drive arm, hand, and f<strong>in</strong>ger movements<br />

are modulated by oscillations <strong>in</strong> <strong>the</strong> delta, <strong>the</strong>ta, mu,<br />

and beta bands (McAuley et al., 1999a; Pfurtscheller<br />

et al., 2000; P<strong>in</strong>eda, 2005; Birbaumer et al., 2006;<br />

Wolpaw, 2007; Hatsopoulos and Donoghue, 2009),<br />

rhythm is imposed on somatosensory <strong>in</strong>flow. (See <strong>the</strong><br />

chapter Rhythms <strong>in</strong> Motor Process<strong>in</strong>g: Functional<br />

Implications for Motor Behavior by N. Hatsopoulos,<br />

<strong>in</strong> this short course.) In <strong>the</strong> visual doma<strong>in</strong>, many<br />

stimuli considered “biologically significant” stem<br />

from observ<strong>in</strong>g <strong>the</strong> motion of conspecifics or o<strong>the</strong>r<br />

animals; here too, because of <strong>the</strong> oscillatory pattern<strong>in</strong>g<br />

mechanisms (above), <strong>the</strong> sensory <strong>in</strong>flow has a strong<br />

rhythmic component. Although a visual scene may<br />

be completely static, <strong>the</strong> sequence of fixations used<br />

to encode <strong>the</strong> scene is generally rhythmic, with a rate<br />

of ~3 Hz (McAuley et al., 1999b; Otero-Millan et<br />

al., 2008). Even when a subject is fixated on a static<br />

stimulus, microsaccades (also occurr<strong>in</strong>g at ~3 Hz)<br />

© 2009 Schroeder<br />

<strong>Align<strong>in</strong>g</strong> <strong>the</strong> <strong>Bra<strong>in</strong></strong> <strong>in</strong> a <strong>Rhythmic</strong> <strong>World</strong><br />

impose a strong rhythm on visual process<strong>in</strong>g (Bosman<br />

et al., 2009).<br />

Interaction of neuronal oscillations<br />

and <strong>in</strong>put rhythms<br />

The strik<strong>in</strong>g similarity between <strong>the</strong> bra<strong>in</strong>’s oscillatory<br />

rhythms and those <strong>in</strong> <strong>the</strong> sensorium is unlikely to<br />

be an accident. Several o<strong>the</strong>r presentations <strong>in</strong> this<br />

short course will deal with <strong>the</strong> precise physiological<br />

mechanisms that generate neuronal oscillations<br />

(see chapters Model<strong>in</strong>g Rhythms: from Physiology<br />

to Function by N. Kopell, and Model<strong>in</strong>g Rhythms:<br />

Detailed Cellular Mechanisms of In Vitro Oscillations,<br />

with Emphasis on Very Fast Oscillations by<br />

R. Traub). This chapter will explore <strong>the</strong> relationship<br />

between neuronal oscillations and environmental<br />

rhythms. We will consider <strong>the</strong> manner <strong>in</strong> which <strong>the</strong><br />

bra<strong>in</strong> may use neuronal oscillations as <strong>in</strong>struments<br />

of adaptive bra<strong>in</strong> operations, by align<strong>in</strong>g <strong>the</strong>m with<br />

relevant environmental rhythms (Schroeder and<br />

Lakatos, 2009a).<br />

Oscillations Control<br />

Neuronal Excitability<br />

Current source density (CSD) analysis of local field<br />

potential (LFP) distributions across cortical layers<br />

shows that fluctuations or “oscillations” of voltage<br />

<strong>in</strong> <strong>the</strong> extracellular medium reflect rhythmic,<br />

synchronous alternation of <strong>in</strong>ward and outward<br />

transmembrane current flow <strong>in</strong> <strong>the</strong> local neuronal<br />

ensembles (Lakatos et al., 2005). In agreement with<br />

Bishop’s fundamental proposition (Bishop, 1933),<br />

analysis of concomitant local neuronal fir<strong>in</strong>g (lower<br />

trace multiunit activity [MUA]) <strong>in</strong>dicates that this<br />

current flow alternation reflects a shift between<br />

net depolarized and hyperpolarized states <strong>in</strong> <strong>the</strong><br />

local neuronal ensemble (Lakatos et al., 2005).<br />

Systematic relationships between oscillatory phase<br />

and excitability have been substantiated for <strong>the</strong> very<br />

low (


NoTeS<br />

24<br />

oscillations as an <strong>in</strong>strument for enhanc<strong>in</strong>g auditory<br />

process<strong>in</strong>g. That is, <strong>the</strong> somatosensory <strong>in</strong>put does not<br />

appear to <strong>in</strong>crease <strong>the</strong> power of ongo<strong>in</strong>g oscillations;<br />

however, it does <strong>in</strong>crease phase coherence across<br />

trials (<strong>in</strong>tertrial coherence, or ITC) <strong>in</strong> <strong>the</strong> low<br />

delta, <strong>the</strong>ta, and gamma bands. The somatosensory<br />

(modulatory) <strong>in</strong>put contrasts with <strong>the</strong> auditory<br />

(driv<strong>in</strong>g) <strong>in</strong>put, which provokes both <strong>in</strong>crease <strong>in</strong><br />

oscillatory power and ITC across <strong>the</strong> spectrum,<br />

typical of an “evoked” response (Makeig et al., 2004;<br />

Shah et al., 2004). Prestimulus-to-poststimulus<br />

<strong>in</strong>crease <strong>in</strong> phase concentration (<strong>in</strong> <strong>the</strong> absence of<br />

an accompany<strong>in</strong>g <strong>in</strong>crease <strong>in</strong> power) <strong>in</strong>dicates that,<br />

<strong>in</strong> A1, somatosensory <strong>in</strong>put has an effect ma<strong>in</strong>ly by<br />

resett<strong>in</strong>g <strong>the</strong> phase of ongo<strong>in</strong>g oscillations. Thus,<br />

somatosensory <strong>in</strong>put “modulates” <strong>the</strong> dynamics of<br />

activity <strong>in</strong> A1 but does not cause auditory neurons<br />

to respond (Lakatos et al., 2007). Both Ghazanfar et<br />

al. (2005) and Kayser et al. (2008) found that visual<br />

<strong>in</strong>put has similar effects on auditory process<strong>in</strong>g <strong>in</strong><br />

primary and secondary auditory cortical areas. Thus,<br />

it seems that this is a general mechanism whereby<br />

nonpreferred (modulatory) stimuli can affect specific<br />

stimulus process<strong>in</strong>g.<br />

Attentional selection<br />

Recent f<strong>in</strong>d<strong>in</strong>gs show that when behaviorally relevant<br />

stimuli occur <strong>in</strong> predictable (rhythmic) streams, lowfrequency<br />

oscillations can function as <strong>in</strong>struments of<br />

attentional selection <strong>in</strong> primary visual cortex (V1)<br />

(Lakatos et al., 2008). For <strong>the</strong>se studies, we tra<strong>in</strong>ed<br />

monkeys to perform an <strong>in</strong>termodal selection task <strong>in</strong><br />

which auditory and visual stimuli (beeps and flashes)<br />

were delivered <strong>in</strong> jittered but rhythmic <strong>in</strong>terdigitated<br />

streams. In alternate trial blocks, <strong>the</strong> monkey<br />

had to attend to ei<strong>the</strong>r <strong>the</strong> visual or <strong>the</strong> auditory<br />

stimulus stream, and make a manual response to an<br />

<strong>in</strong>frequently presented “oddball” stimulus. Although<br />

this paradigm is artificial, it comb<strong>in</strong>es <strong>the</strong> rhythmic<br />

structure and variability characteristic of many<br />

natural event patterns. The key f<strong>in</strong>d<strong>in</strong>g of this<br />

experiment is that, whereas activity <strong>in</strong> <strong>the</strong> thalamic<br />

<strong>in</strong>put (granular) layers entra<strong>in</strong>s (oscillatorily phaselocks)<br />

reliably to visual <strong>in</strong>put, activity <strong>in</strong> <strong>the</strong><br />

extragranular layers entra<strong>in</strong>s to <strong>the</strong> attended <strong>in</strong>put<br />

stream, whe<strong>the</strong>r visual or auditory. The difference is<br />

most apparent <strong>in</strong> <strong>the</strong> supragranular lam<strong>in</strong>ae: A direct<br />

comparison of CSD responses across <strong>the</strong> attend and<br />

ignore conditions shows that delta oscillations <strong>in</strong><br />

<strong>the</strong> supragranular layers are entra<strong>in</strong>ed <strong>in</strong> opposite<br />

phase <strong>in</strong> <strong>the</strong> two attention conditions. The same<br />

comparison for <strong>the</strong> granular layers shows amplitude,<br />

but not phase modulation by attention.<br />

These effects share several complexities, <strong>in</strong>clud<strong>in</strong>g<br />

concomitant <strong>the</strong>ta and gamma band modulations<br />

(Lakatos et al., 2008). However, <strong>the</strong> key factor is<br />

<strong>the</strong> relationship between delta oscillation phase and<br />

neuronal excitability and <strong>the</strong> modulation (shift<strong>in</strong>g)<br />

of delta phase by attention. In broad terms: (1) In<br />

<strong>the</strong> attend visual condition, <strong>the</strong> lam<strong>in</strong>ar CSD profile<br />

at stimulus onset reflects a high excitability state<br />

that permits <strong>the</strong> transmission of <strong>in</strong>puts from granular<br />

to extragranular lam<strong>in</strong>ae (and onward), whereas <strong>in</strong><br />

<strong>the</strong> ignore visual (attend auditory) condition, <strong>the</strong><br />

lam<strong>in</strong>ar CSD configuration at this time po<strong>in</strong>t reflects<br />

a relative depression of excitability <strong>in</strong> this circuit.<br />

(2) These differential facilitative and suppressive<br />

states are responsible for attention’s effects on visual<br />

response amplitudes.<br />

Operational mode of <strong>the</strong> bra<strong>in</strong><br />

aligns to task rhythmicity<br />

We have proposed that, depend<strong>in</strong>g on task<br />

demands, <strong>the</strong> bra<strong>in</strong> dynamically shifts between<br />

variably rhythmic operat<strong>in</strong>g modes (Schroeder<br />

and Lakatos, 2009a). At <strong>the</strong> extremes are <strong>the</strong><br />

“cont<strong>in</strong>uous/vigilance” mode and <strong>the</strong> “rhythmic”<br />

mode. The former may be best exemplified by <strong>the</strong><br />

classic vigilance paradigm used <strong>in</strong> Psychology. This<br />

paradigm entails a task <strong>in</strong> which <strong>the</strong> target stimuli<br />

occur at a completely random and unpredictable<br />

time; an example is when a cat watches a mouse<br />

hole, wait<strong>in</strong>g for <strong>the</strong> mouse to appear. However, as<br />

discussed above, a great deal of natural stimulation<br />

is explicitly rhythmic and predictable. Under <strong>the</strong>se<br />

conditions, neuronal oscillations can entra<strong>in</strong> to<br />

<strong>the</strong> structure of <strong>the</strong> stimulus stream and become<br />

<strong>in</strong>strumental <strong>in</strong> sensory process<strong>in</strong>g (Schroeder et<br />

al., 2008) (<strong>the</strong> <strong>in</strong>termodal paradigm described above<br />

is of this type). <strong>Rhythmic</strong> mode operation entails<br />

entra<strong>in</strong><strong>in</strong>g activity to <strong>the</strong> temporal structure of an<br />

attended stream, result<strong>in</strong>g <strong>in</strong> <strong>the</strong> alignment of highexcitability<br />

oscillation phases with events <strong>in</strong> <strong>the</strong><br />

attended stream. <strong>Rhythmic</strong> mode entra<strong>in</strong>ment, <strong>in</strong><br />

turn, leads to systematic enhancement of responses<br />

to attended events, and suppression of responses to<br />

events that occur out of phase with attended events.<br />

In contrast, when <strong>the</strong>re is no task-relevant rhythm<br />

to which <strong>the</strong> system can entra<strong>in</strong>, low-frequency<br />

oscillations tend to blunt neuronal process<strong>in</strong>g, because<br />

<strong>the</strong>y entail relatively long periods of low excitability<br />

dur<strong>in</strong>g which detection of subtle stimuli would be<br />

impaired. Under <strong>the</strong>se conditions, a cont<strong>in</strong>uous<br />

(vigilance) mode of operation is implemented, lowfrequency<br />

oscillations are suppressed, and <strong>the</strong> system<br />

© 2009 Schroeder


is pushed as far as possible <strong>in</strong>to a cont<strong>in</strong>uous state of<br />

high excitability.<br />

Several behavioral observations are consistent with<br />

<strong>the</strong> differential operation of <strong>the</strong>se two process<strong>in</strong>g<br />

modes. In cont<strong>in</strong>uous/vigilance mode, where <strong>the</strong><br />

oscillatory correlates of attention are enhanced<br />

gamma amplitude/synchrony and lower frequency<br />

suppression (Fries et al., 2001), variations <strong>in</strong> gamma<br />

synchrony are predictive of reaction time variations<br />

(Womelsdorf et al., 2006). In rhythmic mode, where<br />

attention uses low-frequency rhythmic entra<strong>in</strong>ment<br />

(Lakatos et al., 2008), low-frequency phase predicts<br />

reaction time.<br />

We have suggested (Schroeder and Lakatos, 2009a)<br />

that rhythmic mode is <strong>the</strong> preferred state of <strong>the</strong><br />

system for <strong>the</strong> follow<strong>in</strong>g reasons:<br />

(1) It is efficient because <strong>in</strong>puts that are out of phase<br />

with <strong>the</strong> attended stream are automatically<br />

suppressed;<br />

(2) Gamma-band activity appears to be more<br />

metabolically demand<strong>in</strong>g than low-frequency<br />

oscillations (Mukamel et al., 2005; Niess<strong>in</strong>g et<br />

al., 2005); and<br />

(3) Ow<strong>in</strong>g to hierarchical coupl<strong>in</strong>g, gamma activity<br />

is “rationed” (selectively enhanced) at critical<br />

time po<strong>in</strong>ts, when a high-excitability state is<br />

most useful. In this sense, <strong>the</strong> gamma oscillation<br />

is a “slave” ra<strong>the</strong>r than a “master” operator<br />

(Schroeder and Lakatos, 2009b).<br />

Converg<strong>in</strong>g Theory and Reflection<br />

on Earlier F<strong>in</strong>d<strong>in</strong>gs<br />

“Dynamic attend<strong>in</strong>g <strong>the</strong>ory” proposes to describe<br />

many key components of rhythmic mode process<strong>in</strong>g<br />

from a psychophysical perspective (Jones and Boltz,<br />

1989; Large and Jones, 1999; Jones et al., 2006). This<br />

idea has also been developed <strong>in</strong> a motor framework<br />

(Praamstra et al., 2006). The basic hypo<strong>the</strong>sis put<br />

forward by Jones and colleagues is that attend<strong>in</strong>g<br />

itself can be an oscillatory process that entra<strong>in</strong>s<br />

to environmental rhythms, <strong>the</strong>reby improv<strong>in</strong>g<br />

discrim<strong>in</strong>ative performance. In a similar ve<strong>in</strong>, Nobre<br />

and colleagues have suggested that “attention to<br />

time” is a fundamental form of attend<strong>in</strong>g, controlled<br />

by a parietocentric network of bra<strong>in</strong> structures (Nobre<br />

et al., 2007; Correa and Nobre, 2008a,b). There<br />

is evidence (Ghose and Maunsell, 2002) that <strong>in</strong> a<br />

temporally structured (rhythmic) task, monkeys form<br />

an <strong>in</strong>ternal representation of task tim<strong>in</strong>g that can<br />

guide <strong>the</strong> temporal allocation of attentional resources<br />

<strong>in</strong> order to maximize behavioral performance. In all<br />

© 2009 Schroeder<br />

<strong>Align<strong>in</strong>g</strong> <strong>the</strong> <strong>Bra<strong>in</strong></strong> <strong>in</strong> a <strong>Rhythmic</strong> <strong>World</strong><br />

<strong>the</strong>se cases, entra<strong>in</strong>ed neuronal oscillations, operat<strong>in</strong>g<br />

<strong>in</strong> a steady-state mode and/or resett<strong>in</strong>g on a trial-bytrial<br />

basis, provide likely physiological substrates for<br />

<strong>the</strong> effects of attention. The fact that, because of<br />

cross-frequency coupl<strong>in</strong>g, oscillations can be reset to<br />

low-excitability as well as high-excitability states on<br />

multiple time scales (Schroeder et al., 2008) <strong>in</strong>creases<br />

<strong>the</strong> flexibility and range of <strong>the</strong> mechanism.<br />

The dist<strong>in</strong>ction between rhythmic and cont<strong>in</strong>uous<br />

modes of process<strong>in</strong>g may be fundamental: Evidence<br />

of rhythmic mode operation can be seen <strong>in</strong> diverse<br />

varieties of attention, <strong>in</strong>clud<strong>in</strong>g “object attention”<br />

(Taylor et al., 2005), whereas “<strong>in</strong>termodal attention”<br />

(Lakatos et al., 2008) can operate <strong>in</strong> a rhythmic<br />

mode, with excitability/gamma burst<strong>in</strong>g coupled to<br />

<strong>the</strong> rhythm of <strong>the</strong> task. Significantly, <strong>the</strong> most heavily<br />

studied form of attention—spatial attention—<br />

can operate ei<strong>the</strong>r <strong>in</strong> cont<strong>in</strong>uous mode, with<br />

<strong>the</strong> effect of tonic <strong>in</strong>crease <strong>in</strong> excitability (Fries et<br />

al., 2001; Womelsdorf et al., 2006), or <strong>in</strong> rhythmic<br />

mode, with periodic variations <strong>in</strong> excitability coupled<br />

to <strong>the</strong> temporal structure of <strong>the</strong> task (Ghose and<br />

Maunsell, 2002).<br />

The idea that several modes govern <strong>the</strong> operation<br />

of attention may help resolve discrepancies between<br />

attention effects reported by different laboratories<br />

(Moran and Desimone, 1985; Luck et al., 1997, versus<br />

results from McAdams and Maunsell, 1999; Treue and<br />

Maunsell, 1999). Exam<strong>in</strong><strong>in</strong>g <strong>the</strong> task structures used<br />

by <strong>the</strong>se groups suggests that <strong>the</strong> former generally<br />

exam<strong>in</strong>e function<strong>in</strong>g <strong>in</strong> <strong>the</strong> cont<strong>in</strong>uous/vigilance<br />

mode, whereas <strong>the</strong> latter do so <strong>in</strong> <strong>the</strong> rhythmic mode.<br />

Phase-resett<strong>in</strong>g of low-frequency oscillations may help<br />

expla<strong>in</strong> cue<strong>in</strong>g effects on attentional performance<br />

and pert<strong>in</strong>ent event-related potentials (ERPs).<br />

The frontal cont<strong>in</strong>gent negative variation (CNV)<br />

(Walter et al., 1964), for example, may be generated<br />

by phase-resett<strong>in</strong>g frontal low-frequency oscillations<br />

by a warn<strong>in</strong>g cue, which are <strong>the</strong>n averaged over trials.<br />

The perceptual effects known as “attentional bl<strong>in</strong>k”<br />

(Raymond et al., 1992) and “<strong>in</strong>hibition of return,” or<br />

IOR (Kle<strong>in</strong>, 2000), result when stimuli are delivered<br />

dur<strong>in</strong>g <strong>the</strong> low-excitability phase of a low-frequency<br />

oscillation that has been reset by <strong>the</strong> appearance of<br />

ei<strong>the</strong>r a salient target or a cue to attend.<br />

Outstand<strong>in</strong>g questions<br />

Is rhythmic mode process<strong>in</strong>g<br />

generalizable?<br />

A number of empirical predictions need to<br />

be evaluated. Sensory selection <strong>in</strong> a typical<br />

25<br />

NoTeS


NoTeS<br />

26<br />

spatial attention paradigm, for example, could<br />

be accomplished by entra<strong>in</strong><strong>in</strong>g low-frequency<br />

oscillations <strong>in</strong> <strong>the</strong> neuronal representations of <strong>the</strong><br />

“relevant” locations to <strong>the</strong> basic rhythm of stimulus<br />

presentation. The representations of all o<strong>the</strong>r<br />

locations would be left <strong>in</strong> random phase, <strong>the</strong>reby<br />

passively degrad<strong>in</strong>g <strong>the</strong> process<strong>in</strong>g of irrelevant<br />

event streams. Alternatively, activity at sites<br />

represent<strong>in</strong>g irrelevant locations could be aligned <strong>in</strong><br />

counterphase to activity <strong>in</strong> <strong>the</strong> site represent<strong>in</strong>g <strong>the</strong><br />

relevant location, which would produce active (and<br />

stronger) degradation of process<strong>in</strong>g. This technique<br />

would be particularly useful for deal<strong>in</strong>g with locations<br />

conta<strong>in</strong><strong>in</strong>g distractor events. Similar predictions<br />

could be applied to feature-based and object-<br />

based selection.<br />

What is <strong>the</strong> role of corollary eye<br />

movement/fixation signals <strong>in</strong><br />

predictive phase alignment?<br />

Primates, <strong>in</strong>clud<strong>in</strong>g humans, actively exam<strong>in</strong>e <strong>the</strong><br />

visual world by rapidly shift<strong>in</strong>g gaze (fixation) over<br />

<strong>the</strong> elements <strong>in</strong> a scene. Our earlier f<strong>in</strong>d<strong>in</strong>gs (Rajkai<br />

et al., 2007) <strong>in</strong>dicate that a brief period of <strong>in</strong>crease <strong>in</strong><br />

cortical excitability follows each fixation. Just after<br />

fixation onset: (1) The neuronal oscillation phase<br />

transitions from random to a highly organized state;<br />

(2) This phase concentration is accompanied by<br />

<strong>in</strong>creased spectral power <strong>in</strong> several frequency bands;<br />

and (3) Visual response amplitude is enhanced at <strong>the</strong><br />

specific oscillatory phase associated with fixation.<br />

Based on <strong>the</strong>se f<strong>in</strong>d<strong>in</strong>gs, we have hypo<strong>the</strong>sized<br />

that <strong>the</strong> bra<strong>in</strong> uses nonvisual “corollary” signals to<br />

<strong>in</strong>crease cortical excitability at fixation onset, thus<br />

predictively “prim<strong>in</strong>g” <strong>the</strong> system for new visual<br />

<strong>in</strong>puts generated at fixation. Several o<strong>the</strong>r research<br />

groups have endorsed this view (MacEvoy et al.,<br />

2008; Melloni et al., 2009), and thus it appears to<br />

be ga<strong>in</strong><strong>in</strong>g currency. Because <strong>the</strong>se effects have been<br />

identified <strong>in</strong> only a subset of <strong>the</strong> system, it will be<br />

important to determ<strong>in</strong>e how widespread <strong>the</strong>y are. It<br />

will also be important to determ<strong>in</strong>e whe<strong>the</strong>r, as we<br />

suspect, fixation-related phase reset also functions to<br />

<strong>in</strong>crease <strong>the</strong> synchrony of oscillations across layers<br />

with<strong>in</strong> areas, as well as across areas; this determ<strong>in</strong>ation<br />

would clearly contribute to amplification of <strong>the</strong><br />

neuronal representation of <strong>the</strong> object that forms <strong>the</strong><br />

target of fixation.<br />

Go<strong>in</strong>g Forward<br />

Natural stimulation acquired through one’s own<br />

motor behavior, or produced by that of ano<strong>the</strong>r<br />

animal, is generally rhythmic. When <strong>in</strong>tr<strong>in</strong>sic bra<strong>in</strong><br />

oscillations can entra<strong>in</strong> to relevant stimulus rhythms,<br />

<strong>the</strong> bra<strong>in</strong> operates <strong>in</strong> a rhythmic mode, that is,<br />

ambient neuronal oscillations amplify <strong>the</strong> perception<br />

of important <strong>in</strong>puts and suppress perception of<br />

irrelevant ones. When relevant stimuli lack rhythm,<br />

attention operates <strong>in</strong> a cont<strong>in</strong>uous mode, maximiz<strong>in</strong>g<br />

<strong>the</strong> sensitivity of <strong>the</strong> system by suppress<strong>in</strong>g lowerfrequency<br />

oscillations and exploit<strong>in</strong>g <strong>the</strong> advantages<br />

of extended cont<strong>in</strong>uous gamma-band oscillations<br />

(Borgers and Kopell, 2008).<br />

Although questions about rhythmic mode process<strong>in</strong>g<br />

rema<strong>in</strong>, several tentative conclusions are ga<strong>the</strong>r<strong>in</strong>g<br />

weight. Most importantly, prom<strong>in</strong>ent rhythm can<br />

be found <strong>in</strong> many surpris<strong>in</strong>g places. Because of <strong>the</strong><br />

rhythmic <strong>in</strong>fluences of both microsaccades and<br />

macrosaccades (Mart<strong>in</strong>ez-Conde et al., 2004; Rajkai<br />

et al., 2008; Bosman et al., 2009) used to sample a visual<br />

scene, even for static scenes, process<strong>in</strong>g is rhythmic.<br />

The essential rhythmicity of saccades (<strong>in</strong>clud<strong>in</strong>g<br />

microsaccades) has prompted <strong>the</strong> suggestion that<br />

saccadic behavior reflects “an optimal sampl<strong>in</strong>g<br />

strategy by which <strong>the</strong> bra<strong>in</strong> discretely acquires visual<br />

<strong>in</strong>formation” (Otero-Millan et al., 2008b). This<br />

idea offers strong support to that of a characteristic<br />

“attentional sampl<strong>in</strong>g rate” (VanRullen et al., 2007).<br />

Both of <strong>the</strong>se notions re<strong>in</strong>force <strong>the</strong> <strong>the</strong>ory that lowfrequency<br />

rhythms are <strong>in</strong>strumental ra<strong>the</strong>r than<br />

<strong>in</strong>cidental to bra<strong>in</strong> operations.<br />

A strong <strong>in</strong>terdependence between bra<strong>in</strong> and<br />

environmental rhythms, dur<strong>in</strong>g both evolution<br />

and ontogeny, is suggested by <strong>the</strong> fact that certa<strong>in</strong><br />

frequencies dom<strong>in</strong>ate <strong>the</strong> spectra of spontaneous<br />

oscillations <strong>in</strong> sensory cortices (Lakatos et al., 2005)<br />

as well as by <strong>the</strong> remarkable match between <strong>the</strong>se<br />

oscillatory bands and <strong>the</strong> temporal structure of<br />

biologically relevant sensory <strong>in</strong>puts (Lou and Poeppel,<br />

2007; Schroeder et al., 2008). Widespread crossfrequency,<br />

oscillatory coupl<strong>in</strong>g offers <strong>the</strong> potential<br />

for improv<strong>in</strong>g <strong>the</strong> match<strong>in</strong>g of bra<strong>in</strong> oscillations with<br />

complex natural <strong>in</strong>put patterns that cover different<br />

time scales (Schroeder et al., 2008), as well as more<br />

subtle computational benefits (Krupa et al., 2008).<br />

The modulation of low-frequency phase, and of<br />

cross-frequency coupl<strong>in</strong>g by attention (Lakatos et al.,<br />

2008) molds sensory process<strong>in</strong>g to <strong>the</strong> current goals<br />

of an observer.<br />

References<br />

Birbaumer N, Weber C, Neuper C, Buch E, Haapen<br />

K, Cohen L (2006) Physiological regulation of<br />

th<strong>in</strong>k<strong>in</strong>g: bra<strong>in</strong>-computer <strong>in</strong>terface (BCI) research.<br />

Prog <strong>Bra<strong>in</strong></strong> Res 159:369-391.<br />

Bishop G (1933) Cyclical changes <strong>in</strong> excitability<br />

of <strong>the</strong> optic pathway of <strong>the</strong> rabbit. Am J Physiol<br />

103:213-224.<br />

© 2009 Schroeder


Borgers C, Kopell NJ (2008) Gamma oscillations and<br />

stimulus selection. Neural Comput 20:383-414.<br />

Bosman C, Womelsdorf T, Desimone R, Fries P<br />

(2009) A microsaccadic rhythm modulates gammaband<br />

synchronization and behavior. J Neurosci<br />

29:9471-9480.<br />

Buzsaki G (2007) The structure of consciousness.<br />

Nature 446:267.<br />

Buzsaki G, Draguhn A (2004) Neuronal oscillations<br />

<strong>in</strong> cortical networks. Science 304:1926-1929.<br />

Contreras D, Timofeev I, Steriade M (1996)<br />

Mechanisms of long-last<strong>in</strong>g hyperpolarizations<br />

underly<strong>in</strong>g slow sleep oscillations <strong>in</strong> cat<br />

corticothalamic networks. J Physiol 494 (Pt<br />

1):251-264.<br />

Correa A, Nobre AC (2008a) Neural modulation<br />

by regularity and passage of time. J Neurophysiol<br />

100:1649-1655.<br />

Correa A, Nobre AC (2008b) Spatial and temporal<br />

acuity of visual perception can be enhanced<br />

selectively by attentional set. Exp <strong>Bra<strong>in</strong></strong> Res<br />

189:339-334.<br />

Fries P, Reynolds JH, Rorie AE, Desimone R<br />

(2001) Modulation of oscillatory neuronal<br />

synchronization by selective visual attention.<br />

Science 291:1560-1563.<br />

Fries P, Schroder JH, Roelfsema PR, S<strong>in</strong>ger W, Engel<br />

AK (2002) Oscillatory neuronal synchronization<br />

<strong>in</strong> primary visual cortex as a correlate of stimulus<br />

selection. J Neurosci 22:3739-3754.<br />

Ghazanfar AA, Maier JX, Hoffman KL, Logo<strong>the</strong>tis<br />

NK (2005) Multisensory <strong>in</strong>tegration of dynamic<br />

faces and voices <strong>in</strong> rhesus monkey auditory cortex.<br />

J Neurosci 25:5004-5012.<br />

Ghose GM, Maunsell JH (2002) Attentional<br />

modulation <strong>in</strong> visual cortex depends on task<br />

tim<strong>in</strong>g. Nature 419:616-620.<br />

Hatsopoulos NG, Donoghue JP (2009) The science<br />

of neural <strong>in</strong>terface systems. Annu Rev Neurosci<br />

32:249-266.<br />

Jensen O, Kaiser J, Lachaux JP (2007) Human gammafrequency<br />

oscillations associated with attention<br />

and memory. Trends Neuroscim 30:317-324.<br />

Jones MR, Boltz M (1989) Dynamic attend<strong>in</strong>g and<br />

responses to time. Psychol Rev 96:459-491.<br />

Jones MR, Johnston HM, Puente J (2006) Effects<br />

of auditory pattern structure on anticipatory and<br />

reactive attend<strong>in</strong>g. Cogn Psychol 53:59-96.<br />

Kayser C, Petkov CI, Logo<strong>the</strong>tis NK (2008) Visual<br />

modulation of neurons <strong>in</strong> auditory cortex. Cereb<br />

Cortex 18:1560-1574.<br />

© 2009 Schroeder<br />

<strong>Align<strong>in</strong>g</strong> <strong>the</strong> <strong>Bra<strong>in</strong></strong> <strong>in</strong> a <strong>Rhythmic</strong> <strong>World</strong><br />

Kle<strong>in</strong> RM (2000) Inhibition of return. Trends Cogn<br />

Sci 4:138-147.<br />

Krupa M, Popovic N, Kopell N, Rotste<strong>in</strong> HG<br />

(2008) Mixed-mode oscillations <strong>in</strong> a three timescale<br />

model for <strong>the</strong> dopam<strong>in</strong>ergic neuron. Chaos<br />

18:015106.<br />

Lakatos P, Shah AS, Knuth KH, Ulbert I, Karmos<br />

G, Schroeder CE (2005) An oscillatory hierarchy<br />

controll<strong>in</strong>g neuronal excitability and stimulus<br />

process<strong>in</strong>g <strong>in</strong> <strong>the</strong> auditory cortex. J Neurophysiol<br />

94:1904-1911.<br />

Lakatos P, Chen CM, O’Connell MN, Mills A,<br />

Schroeder CE (2007) Neuronal oscillations and<br />

multisensory <strong>in</strong>teraction <strong>in</strong> primary auditory<br />

cortex. Neuron 53:1-14.<br />

Lakatos P, Karmos G, Mehta AD, Ulbert I, Schroeder<br />

CE (2008) Oscillatory entra<strong>in</strong>ment as a mechanism<br />

of attentional selection. Science 320:110-113.<br />

Large EW, Jones MR (1999) The dynamics of<br />

attend<strong>in</strong>g: how people track time-vary<strong>in</strong>g events.<br />

Psychol Rev 30:119-159.<br />

Lou H, Poeppel D (2007) Phase patterns of neuronal<br />

responses reliably discrim<strong>in</strong>ate speech <strong>in</strong> human<br />

auditory cortex. Neuron 54:1001-1010.<br />

Luck SJ, Chelazzi L, Hillyard SA, Desimone R (1997)<br />

Neural mechanisms of spatial selective attention<br />

<strong>in</strong> areas V1, V2 and V4 of macaque visual cortex.<br />

J Neurophysiol 77:24-42.<br />

MacEvoy SP, Hanks TD, Paradiso MA (2008)<br />

Macaque V1 activity dur<strong>in</strong>g natural vision: effects<br />

of natural scenes and saccades. J Neurophysiol<br />

99:460-472.<br />

Makeig S, Debener S, Onton J, Delorme A (2004)<br />

M<strong>in</strong><strong>in</strong>g event-related bra<strong>in</strong> dynamics. Trends<br />

Cogn Sci 8:204-210.<br />

Mart<strong>in</strong>ez-Conde S, Macknik SL, Hubel DH (2004)<br />

The role of fixational eye movements <strong>in</strong> visual<br />

perception. Nat Rev Neurosci 5:229-240.<br />

Maurer AP, McNaughten BL (2007) Network and<br />

<strong>in</strong>tr<strong>in</strong>sic cellular mechanisms underly<strong>in</strong>g <strong>the</strong>ta<br />

phase precession of hippocampal neurons. Trends<br />

Neurosci 30:325-333.<br />

McAdams CJ, Maunsell JHR (1999) Effects of<br />

attention on orientation-tun<strong>in</strong>g functions of s<strong>in</strong>gle<br />

neurons <strong>in</strong> macaque cortical area V4. J Neurosci<br />

19:431-441.<br />

McAuley JH, Farmer SF, Rothwell JC, Marsden<br />

CD (1999a) Common 3 and 10 Hz oscillations<br />

modulate human eye and f<strong>in</strong>ger movements while<br />

<strong>the</strong>y simultaneously track a visual target. J Physiol<br />

515(Pt 3):905-917.<br />

27<br />

NoTeS


NoTeS<br />

28<br />

McAuley JH, Rothwell JC, Marsden CD (1999b)<br />

Human anticipatory eye movements may reflect<br />

rhythmic central nervous activity. Neuroscience<br />

94:339-350.<br />

Melloni L, Schwiedrzik CM, Rodriguez E, S<strong>in</strong>ger<br />

W (2009) (Micro)saccades, corollary activity and<br />

cortical oscillations. Trends Cogn Sci 13:239-245.<br />

Moran J, Desimone R (1985) Selective attention<br />

gates visual <strong>in</strong>formation process<strong>in</strong>g <strong>in</strong> extrastriate<br />

cortex. Science 229:782-784.<br />

Mukamel R, Gelbard H, Arieli A, Hasson U, Fried I,<br />

Malach R (2005) Coupl<strong>in</strong>g between neuronal fir<strong>in</strong>g,<br />

field potentials, and FMRI <strong>in</strong> human auditory cortex.<br />

Science 309:951-954.<br />

Niess<strong>in</strong>g J, Ebisch B, Schmidt KE, Niess<strong>in</strong>g M,<br />

S<strong>in</strong>ger W, Galuske RA (2005) Hemodynamic<br />

signals correlate tightly with synchronized gamma<br />

oscillations. Science 309:948-951.<br />

Nobre A, Correa A, Coull J (2007) The hazards of<br />

time. Curr Op<strong>in</strong> Neurobiol 17:465-470.<br />

Otero-Millan J, Troncoso XG, Macknik SL, Serrano-<br />

Pedraza I, Mart<strong>in</strong>ez-Conde S (2008) Saccades and<br />

microsaccades dur<strong>in</strong>g visual fixation, exploration,<br />

and search: foundations for a common saccadic<br />

generator. J Vis 8:21.1-21.18.<br />

Palva S, Palva JM (2008) New vistas for alphafrequency<br />

band oscillations. Trends Neurosci<br />

30:150-158.<br />

Pfurtscheller G, Neuper C, Kraut M (2000) Functional<br />

dissociation of lower and upper frequency mu<br />

rhythms <strong>in</strong> relation to voluntary limb movement.<br />

Cl<strong>in</strong> Neurophysiol 111:1873-1879.<br />

P<strong>in</strong>eda JA (2005) The functional significance of mu<br />

rhythms: translat<strong>in</strong>g “see<strong>in</strong>g” and “hear<strong>in</strong>g” <strong>in</strong>to<br />

“do<strong>in</strong>g”. <strong>Bra<strong>in</strong></strong> Res <strong>Bra<strong>in</strong></strong> Res Rev 50:57-68.<br />

Praamstra P, Kourtis D, Kwok HF, Oostenveld R<br />

(2006) Neurophysiology of implicit tim<strong>in</strong>g <strong>in</strong> serial<br />

choice reaction-time performance. J Neurosci<br />

26:5448-5455.<br />

Rajkai C, Lakatos P, Chen C, P<strong>in</strong>cze Z, Karmos G,<br />

Schroeder CE (2007) Transient cortical excitation<br />

at <strong>the</strong> onset of visual fixation. Cereb Cortex<br />

18:200-209.<br />

Rajkai C, Lakatos P, Schroeder CE (2008) Visual<br />

fixation-related activity <strong>in</strong> <strong>the</strong> auditory cortex of<br />

monkeys.<br />

Raymond JE, Shapiro KL, Arnell KM (1992)<br />

Temporary suppression of visual process<strong>in</strong>g <strong>in</strong> an<br />

RSVP task: an attentional bl<strong>in</strong>k? J Exp Psychol<br />

Hum Percept Perform 18:849-860.<br />

Sanchez-Vives MV, McCormick DA (2000) Cellular<br />

and network mechanisms of rhythmic recurrent<br />

activity <strong>in</strong> neocortex. Nat Neurosci 3:1027-1034.<br />

Schroeder CE, Lakatos P (2009a) Low-frequency<br />

neuronal oscillations as <strong>in</strong>struments of sensory<br />

selection. Trends Neurosci 32:9-18.<br />

Schroeder CE, Lakatos P (2009b) The gamma<br />

oscillation: master or slave? <strong>Bra<strong>in</strong></strong> Topogr<br />

22:24-26.<br />

Schroeder CE, Lakatos P, Kajikawa Y, Partan S,<br />

Puce A (2008) Neuronal oscillations and<br />

visual amplification of speech. Trends Cogn Sci<br />

12:106-113.<br />

Shah AS, Bressler SL, Knuth KH, D<strong>in</strong>g M, Mehta<br />

AD, Schroeder CE (2004) Neural dynamics<br />

and fundamental mechanisms of event-related<br />

potentials. Cereb Cortex 14:476-485.<br />

Sherman SM, Guillery RW (2002) The role of<br />

<strong>the</strong> thalamus <strong>in</strong> <strong>the</strong> flow of <strong>in</strong>formation to <strong>the</strong><br />

cortex. Philos Trans R Soc Lond B Biol Sci<br />

357:1695-1708.<br />

S<strong>in</strong>ger W, Gray CM (1995) Visual feature <strong>in</strong>tegration<br />

and <strong>the</strong> temporal correlation hypo<strong>the</strong>sis. Annu<br />

Rev Neurosci 18:555-586.<br />

Steriade M, Nunez A, Amzica F (1993) A novel slow<br />

(

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!