this new draft by Ramirez, Theofanopoulou and Boeckx

A hypothesis concerning the neurobiological basis
of phrase structure building∗
Javier Ram´ırez1 , Constantina Theofanopoulou2 & Cedric Boeckx2,3
Universitat de Girona,
Universitat de Barcelona,
correspondence: [email protected]
March 2015
The goal of this paper is to offer a theoretical model of how the aspect of our
linguistic cognition known as phrase-structure building may be implemented in
neural terms. Our proposal concerning this brain implementation requires us
to decompose phrase-structure building into more elementary sub-processes,
which we suggest can be captured in terms of brain rhythms. The rhythms
involved—alpha, beta, gamma, and theta—are generated in various cortical
and sub-cortical structures, and form part of a widely-distributed network
that we argue is superior to more classical approaches in neurolinguistics that
rely mostly on a fronto-temporal circuit.
keywords: phrase structure, oscillation, syntax, frequency-coupling, language.
It is beyond dispute that members of our species, barring the most severe
pathologies or environmental circumstances, have the cognitive capacity to
We are indebted to David Poeppel for invaluable comments and advice. This work was made
possible through a Marie Curie International Reintegration Grant from the European Union (PIRGGA-2009-256413), research funds from the Fundaci´o Bosch i Gimpera, a grant from the Spanish
Ministry of Economy and Competitiveness (FFI-2013-43823-P), a grant from the Generalitat de
Catalunya (2014-SGR-200), and an FI-fellowship from the Generalitat de Catalunya.
acquire grammatical systems of great internal complexity. At the heart of this
capacity lies a process that one can call “Phrase Structure building”. This
process takes elementary units as its input and organizes them into nested
sets. It is this process that leads to the dictum that words in sentences are not
mere beads on a string. A major challenge for us is that unlike phonological
representations, which are linked to perceptual units, syntactic representations
rely on mind-internal, self-generated cognitive inputs.
Over thirty years of intensive research on a wide range of languages have
made it clear that this Phrase Structure Building capacity (hereafter, PSBC) is
not subject to variation (Boeckx, 2014). It underlies the construction of every
sentence in every language, even if each language exhibits distinct morphophonological properties that “dress” phrases in many ways, thereby providing
the major source of linguistic diversity. Put another way, PSBC can to a significant extent be kept separate from cross-linguistic considerations pertaining
to variation and change. One can even go further and argue that traditional
grammatical conceptions that take language-specific units like “words” to provide instructions to build phrases are wrong, and that instead, lexical items
are the expressions or outputs of syntactic structures (Hale and Keyser, 1993,
Halle and Marantz, 1993, Boeckx, 2014). In others words, PSBC provides the
scaffolding for language-specific morphological elaborations.
The goal of this paper is to offer a theoretical model of how this invariant
aspect of our grammar-forming capacity may be implemented in brain terms.
We have chosen to focus on PSBC, as opposed to more general notions such
as ‘syntax’ because along with Ben Shalom and Poeppel (2008) we think that
traditional research areas in linguistics should not be expected to map neatly
onto brain units, be they areas, regions, or elementary networks. Rather, the
focus should be on processes that ultimately can be decomposed into primitive
and generic operations of the sort the brain is known to perform. Thus, our
approach to ‘language’ fully agrees with the ‘divide-and-conquer’ perspective
advocated by Schaafsma, Pfaff, Spunt, and Adolphs (2015) for other cognitive
traits like “Theory of Mind”. These authors are right to stress that standard
cognitive units do not permit easy downward translation to more basic processes such as those studied in neuroscience, and if taken as starting points only
serve to exacerbate the granularity mismatch problem highlighted by Poeppel
and Embick (2005), Embick and Poeppel (2014). The divide-and-conquer approach has recently yielded significant results in evolutionary studies grounded
in comparative cognitive considerations (Ravignani et al., 2014), and we think
that it could prove equally fruitful in bridging the gap between mind and brain.
In addition, the divide-and-conquer approach aligns well with recent attempts
in theoretical linguistics to approach components of the human language faculty “from below” (Chomsky, 2007), which also facilitates the formulation of
linking hypotheses between fields. An immediate consequence of this divideand-conquer approach is that it forces to adopt a new perspective on how to
ground cognition in brain function. In the context of language, this meshes well
with the numerous calls to go beyond the classical Broca-Lichtheim-Wernicke
model (Poeppel et al., 2012, Poeppel, 2014, Fedorenko and Thompson-Schill,
2014, Hagoort, 2014).
Decomposing Phrase Structure Building
Although already more appropriate than notions like ‘syntax’, ‘Phrase Structure Building’ itself is not a single, atomic concept. We claim that at least
4 distinct sub-processes must be distinguished if PSBC is to be mapped onto
brain operations. We should point out at the outset that these sub-processes
receive different names in different theoretical traditions, but we think that
on the whole the recognition of these four characteristics is relatively theoryneutral.
The first sub-process we dub ‘atomization’. It sets the stage for what is
perhaps the most distinctive aspect of linguistic cognition. In principle, any
two “conceptual” units can be combined into a phrase in the linguistic domain.
Of course, in practice, languages impose many restrictions on this combination,
filtering out many unwanted phrases. Likewise, many of these combinations
don’t immediately make a lot of sense (think of colorless green ideas sleep
furiously). But in theory any two units can be paired into a phrase. This is
quite unlike what we find in more ‘primitive’ cognitive domains (e.g., the core
knowledge systems of Spelke and Kinzler (2007)), which impose severe ‘adicity’
restrictions reminiscent of ‘chemical valences’. As has been documented by
numerous experimental and comparative psychologists, non-linguistic minds
are highly modular. In the absence of language, cross-modular combinations
are extremely rare and unstable (Spelke, 2003, Hauser, 2009). Not surprisingly,
several authors have seen in language the mechanism by which cross-modular
thoughts can be formed and robustly used (Spelke, 2003, Carruthers, 2006,
Boeckx, 2011, Pietroski, 2012, Reinhart, 2006, Mithen, 1996). By placing
concepts in the language domain, ‘selectional’ restrictions can be lifted, and
combinatorial possibilities expand dramatically, leading to creativity.
The second sub-process amounts to set-formation, a term we prefer to
arrays or sequences, although we will return to the notion of ‘hierarchical sequences’ at several points in this paper, especially in the context of the chunking role of the basal ganglia. Set-formation is the operation that combines
linguistic units. We will refer to it as ‘combine’ in what follows. Many linguists take this combination process to be restricted to the combination of two
elements (Kayne, 1984, 1994). We will not provide a strong argument in favor
of this position, but will simply adopt it here as the simplest case scenario.
The third sub-process is one of categorization or labeling. It is the process
by which a phrase acquires its identity. Phrases in language come in several
varieties: verb phrases, noun phrases, and so on. Labeling is the operation by
which a particular unit entering the combination process gives an entire phrase
its identity. Traditionally, this is done ‘endocentrically’ or ‘from within’: one
of the two units combined gives the phrase its name (Chomsky, 1970). More
recently, though, it has been pointed out that labeling is more accurately
thought of as an ‘exo-skeletal’ process: a phrase receives its entity based on its
‘entourage’, that is, the elements immediately surrounding it (Marantz, 2008,
Borer, 2005, Boeckx, 2014). A noun phrase is a noun phrase in virtue of being
surrounded by a determiner (the book), not because it contains a noun. A verb
phrase is a verb phrase not because it contains a verb, but because a tense
marker immediately dominates it (to book). Advocates of the exo-skeletal
approach to labeling often point out that this process is transparently at work
in the linguistic interpretation that speakers naturally impose on poems like
Lewis Carroll’s Jabberwocky. Thanks to recognizable units like determiners,
tense markers, etc., we come to identify nouns and verbs that otherwise could
not be driving the labeling process, being as they are novel, unknown, and
therefore a-categorial.
The fourth sub-process is intrinsically tied to the third one. Labeling is
monotonic in that once a phrase has been labeled and is embedded in another
phrase, its identity is preserved. This monotonicity is often captured in linguistics by appealing to the notion of (strict) cyclicity. The process of labeling
takes place as a sliding window of activity traverses the sentence: once categorized, a phrase is “consolidated” or stored as a chunk, and the next as-yet
unlabeled structure becomes the focus of syntactic decision.
3 Phrase Structure Building and Brain Oscillations
Our proposal concerning the brain implementation of the processes giving rises
to phrases will be couched in terms of brain rhythms. It has long been suspected that the rhythmic fluctuations of electrical activity produced by the
brain may play a role in cognition, but in recent years numerous publications
have put forward specific proposals concerning how neural oscillations at different frequencies could be related to a wide range of basic and higher cognitive
processes (Buzs´
aki, 2006). In the language domain perhaps the most detailed
an compelling model is that of Giraud and Poeppel (2012), who show that θ
and γ oscillations are engaged by the multi-timescale, quasi-rhythmic properties of speech and can track its dynamics. In so doing, brain rhythms ‘package’
incoming information into units that provide the basis for representations like
syllables. Outside of language, several authors have shown how brain rhythms
interact with one another to provide the basic computational machinery to
support functions like ‘working memory’ (Lisman, 2005, Dipoppa and Gutkin,
2013, Roux and Uhlhaas, 2014). Buzs´aki (2010) goes so far as to claim that
brain oscillations provide the foundational framework for a neural ‘syntax’, understood as “a set of principles that govern the transformation and temporal
progression of discrete elements into ordered and hierarchical relations”. We
are quite aware that not every issue has been solved in the domain of brain
rhythms, but we feel quite confident that the quantity of works consulted [see
Supplementary Material] provides a solid basis for the claims put forth in this
The rhythms we will implicate to capture the processes that make up PSBC
will be α, β, γ, and θ. We will claim that these rhythms interact in such a way
as to yield the fundamental operations of PSBC. In addition to the cortex,
we will recruit subcortical structures to generate some of these rhythms: the
thalamus for the α band, the basal ganglia for the β band, and the hippocampus for the θ band. The existing literature offers evidence for associating the
generation of these rhythms with these brain structures [see Supplementary
Material], although they rarely assign them a fundamental function in language (but see Theofanopoulou and Boeckx (To appear)). In part this is due
to the fact that neurolinguistics continues to be cortico-centric. But it is also
due to the fact that most neuroscientists think of language in very concrete
terms: words, sounds, etc. Our focus here is on an aspect of language that
is much more closely related to the formation of thoughts, and, as we already
said, is relatively separate from the morpho-phonological substance used in
languages. Not surprisingly, then, the network we envisage depart in significant ways from the traditional fronto-temporal circuit (Friederici and Gierhan,
2013), which we have argued elsewhere (Boeckx et al., 2014) has been wrongly
associated with aspects of phrase structure building (Friederici, 2009), an issue
we return to at the end of this article. Our network also inverts the distinction between language core and periphery (Fedorenko and Thompson-Schill,
2014), attributing to domain-general circuits the role of elementary syntactic
In the cognitive neuroscience literature one finds two large cortical networks that seem to us to relate to the nature of PSBC as we conceive of it:
the Default Mode Network (Raichle et al., 2001) and the Multiple Demand
network (Duncan, 2010). Both consist of distributed cortical regions regulated
in part by subcortical activity. The Default Mode Network (DMN) comprises
portions of the Pre-frontal cortex, the precuneus, the posterior cingulate cortex, the inferior parietal cortex, the lateral temporal cortex. DMN activitity
has been detected for passive or internally-driven tasks, as opposed to active or
externally-driven tasks (Buckner et al., 2008). The Multiple Demand network
(MD) also consists of a fronto-parietal circuit that has been used to capture
cognitive notions like “fluid intelligence” and “complex cognition”. “Mind
wandering” or “divergent (unconventional) thinking” also exploit to Frontoparietal connections, and we think that it is ultimately related to the frontoparietal expansion that gives our species’ brain its distinctive globular shape
(Boeckx and Ben´ıtez-Burraco, 2014). Although often assumed to be anticorrelated networks (Fox et al., 2005), DMN and MD have been shown to be
transiently coupled at the scale of the rhythms we implicate here (de Pasquale
et al., 2012), which we take to confer initial plausibility to our suggestion that
both networks are involved in the model we envisage here. The network we
have in mind also shares properties of the global workspace model in Dehaene
et al. (1998) or the ‘connective core’ in Shanahan (2012).
For purposes of the proposal to come, it is crucial to exploit layer-specific
properties of the neo-cortex, as distinct layers have distinct rhythmic preferences, and thus ultimately play distinct cognitive roles. Here we will rely on
the distinction between supra-granular and infra-granular layers, and, following Bastos et al. (2012), associate high, γ frequency oscillations mostly with
the supra granular layers, and lower frequencies (specifically, β and α) with
infra granular layers.
Infragranular layers connect to higher-order thalamic nuclei such as the
medio-dorsal nucleus and the pulvinar, which generate a robust α rhythm
(Bollimunta et al., 2011) and are crucially involved in cortico-cortical information management (Theyel et al., 2010, Saalmann et al., 2012). Infragranular
layers also connect to the basal ganglia, in particular the striatum, which is
implicated in sequencing and chunking procedures, as well as a process of characterization in tandem with the cortex and higher-order thalamic nuclei like
the medio-dorsal nucleus (Antzoulatos and Miller, 2014). Striatal structures,
as well as the medio-dorsal thalamic nucleus, have been shown to operate
at the β range (Leventhal et al., 2012, Parnaudeau et al., 2013). Finally,
the θ rhythm is the signature rhythm of the hippocampus and associated enthorinal cortex. Through its connections with the thalamus and the basal
ganglia, we hypothesize that the hippocampus is responsible for the theta
rhythm detected in anterior thalamic nuclei and reuniens, and in the ventral
and dorsomedial striatum, respectively. See Figures 1 and 2 for details of the
“connectome”/“dynome” we envisage.
We claim that PBSC arises as the result of the interaction among these
various rhythms. In a nutshell, the α rhythm provides the means to embed
γ activity generated in widely-distributed cortical sources (but also present in
sub-cortical structures such as the basal ganglia), which essentially amounts
to the combination of atomic units across modular cognitive boundaries (suboperations 1 and 2 from section 2). The cyclic consolidation or storage of
partial structures amounts to the periodic embedding of γ cycles inside the
hippocampus-driven θ rhythm (sub-operation 4 from section 2). We hypothesize that for this γ–θ embedding to take place, γ activity must be decoupled
from α activity, a process that is achieved through the action of thalamic
reticular nucleus. Crucially, for this γ–θ pairing to lead to successful cognitive
interpretation, it must be accompanied by a labeling process (sub-operation 3
from section 2), which in oscillatory terms is achieved by the slowing down of
γ activity to the β frequency, and the subsequent coupling of β and α. See
Figure 3 for a schematic representation of our mapping hypothesis.
Let us spell out each claim in what follows, beginning with sub-operation 1
from section 2. Atomization is accomplished by α-embedded neural assemblies
oscillating at γ in supragranular layers of the cortical regions of DMN. This
is consistent with claims in the literature such as the formation of “words” in
“neural syntax” (Buzs´
aki, 2010); the binding of features into coherent objects
(Bosman et al., 2014); the local nature of the operation due to conduction
delays (Von Stein and Sarnthein, 2000); and the directionality of feed-forward
processing attributed to the rhythm (Bastos et al., 2015).
Set-formation (sub-operation 2 from section 2) is accomplished by a crossfrequency coupling mechanism between higher order thalamic nuclei, specifically the dorsal pulvinar, oscillating at α frequency (Saalmann and Kastner,
2011) and the above-described assemblies oscillating at the γ range. The function of the α frequency is consistent with the active processing hypothesis in
Palva and Palva (2011); the working memory literature, where the α rhythm
binds visuo-spatial elements (Roux and Uhlhaas, 2014); the coexistence of α
and γ signals in task-relevant regions of working memory models (Honkanen
et al., 2014) and its modulation of local gamma amplitude (Roux et al., 2013,
Saalmann et al., 2012). The set-formation operation just described could be
regulated by activity of the thalamic reticular nucleus, which would move
higher-order thalamic nuclei such as the pulvinar from a ‘burst” firing mode
that would wake-up the transient cortical nodes of the network (Womelsdorf
et al., 2014), to a second, tonic firing mode that would strengthen the dialog
among the nodes of MD. Crucially for us, we take α activity in higher-order
thalamic nuclei to go beyond the traditional inhibitory function associated
with that frequency in lower-level, perceptual nuclei (Palva and Palva, 2011,
Klimesch, 2012, Theofanopoulou and Boeckx, To appear).
Labeling (sub-operation 3 from section 2) is accomplished by one basal
ganglia-thalamic-cortical loop, likely crossing the dorsolateral striatum, disinhibiting the thalamic medio-dorsal nucleus, by means of the β rhythm, retaining in working memory one of the objects generated by sub-operation 1.
We claim that the basal ganglia select and hold one of the above γ-supported
items, which allows it to stay coupled to the set-forming α rhythm for a longer
period than other γ assemblies momentarily coupled to θ (sub operation 4).
The retained γ rhythm must be slowed down to β frequency due to conduction delays that are unavoidable in the wider neural population recruited in
the model envisaged here. (The distinction we introduce here between the role
of the β-band and the θ-band may well correspond to the function of shortterm vs. working memory, although we leave a detailed examination of this
possibility for the future, as the literature on memory systems is so complex
that we cannot adequately survey it here.)
Such labeling process thus leads us to distinguish two different elements as a
function of the rhythms that sustain them: γ and β objects. This is consistent
with the anatomical and oscillatory boundaries between supra-granular and
infra-granular layers (Bastos et al., 2012) and their direction of information
flow (feed-forward and feedback; Bastos et al. (2015), Miller and Buschman
(2012)). It is also consistent with the different rhythms that sustain objects
as a function of their complexity (Honkanen et al., 2014). There is a sense in
which the system generates two kinds of ‘labels’, as in exoskeletal approaches
in linguistics: a ‘procedural label’ (serving as an instruction to interpret neighboring elements), and a more ‘declarative label’, associated with traditional
categories like nouns, verbs, and adjectives. This procedural/declarative distinction is often associated with the basal ganglia and the hippocampus. In
fact, we hypothesize that this result is achieved when basal ganglia θ and hippocampal θ rhythms get synchronized, as shown in decision periods in various
tasks (Tort et al., 2008, DeCoteau et al., 2007).
Our use of the β frequency is consistent with the notion of top-down control
we assume is required for labeling (Chan et al., 2014), and fits well with the
idea that MD links attentional episodes in a hierarchy (Duncan, 2013). It also
meshes well with the idea that β rhythm is associated with the sustainment of
status quo (Engel and Fries, 2010), as well as with claims in the working memory literature that β holds information (Martin and Ravel, 2014, Engel and
Fries, 2010, Parnaudeau et al., 2013, Tallon-Baudry et al., 2004, Deiber et al.,
2007, Salazar et al., 2012). Ultimately, we think that the β frequency fulfills
the role of non-terminal symbols in early computational models of generative
grammar (Chomsky, 1957).
Sub-operation 4 from section 2 consists of the desynchronization of the
items from the set-forming α rhythm, and their subsequent synchronization
with a θ oscillation generated by the hippocampal complex. Concretely, we
point to the thalamic reticular nucleus as a key element in the desyncrhonization of oscillations by means of local thalamic inhibition, along the lines of
Huguenard and McCormick (2007). We also hypothesize a possible function
of the median raphe nucleus in the final stage of the operation (Vertes et al.,
2004). The periodicity of the sub-operation under discussion is naturally given
by the cross-frequency phase-phase 2:1 coupling of α and θ (Buzs´aki, 2006).
Incidentally, we think that it is this periodicity that gives its very meaning to
the notion of ‘cycle’ attached to linguistic processes.
The involvement of the hippocampus in our model is consistent with its
widely-connected nature (de Pasquale et al., 2012) and claims that it is in
service of other systems (Rubin et al., 2014). Our use of the θ frequency is
consistent with claims concerning its use in the storing of items in working
memory (Roux and Uhlhaas, 2014). The cross-frequency coupling of θ and α
we defend here has already been recognized in the attention literature (Song
et al., 2014), where this coupling controls the multi-item sampling in which θ
dominates alternating periods of α. It may also shed light over the sequential/parallel debate about multi-items with attentional search (Dugu´e et al.,
2014). Indeed, we think that the α rhythm plays an important role in suppressing distracting information during working memory encoding and maintenance, which lies at the heart of the intersection between working memory
and attention.
To a significant extent, PSBC relies on generic, conserved brain rhythm
mechanisms (Buzs´
aki et al., 2013), and are therefore neither language-, nor
human-specific. In our opinion, both species- and cognitive-specificity arise
from the context in which these operations take place. In particular, the
globularity of our species, associated with parietal expansion, certainly led
to an expansion of the pulvinar and a concomitant reduction of the portion
allocated to it that connects with the occipital lobe, itself reduced in anatomically modern humans (Boeckx and Ben´ıtez-Burraco, 2014, Pearce et al., 2013).
Plausibly, this expansion of the pulvinar, along with a more central location of
higher-order nuclei in the brain as a whole, led to a more efficient connectivity
pattern across widely-distributed cortical areas, reducing “spatial inequalities”
(Salami et al., 2003) in other species’ brains that prevent the formation of the
“necessary symmetry” (Bastos et al., 2015) and “temporal equidistance” (Vicente et al., 2008) required to couple rhythms over widely-distributed networks.
In this case, we believe that this anatomical reconfiguration played a key role
in allowing for cross-modular thoughts to be robustly established and put to
Needless to say, the model offered here remains to be tested experimentally.
But we think that in addition to offering a biologically plausible linking hypothesis between cognitive science and neuroscience, it may provide a useful
point of departure to understand cognitive deficits across a wide range of mental disorders, for which we have an increasing amount of information couched
in oscillatory terms (Buzs´
aki and Watson, 2012).
We also believe that some of the grammatical constraints identified in the
theoretical linguistic literature may receive a natural interpretation in terms
of the model presented here. Indeed, the literature on brain frequency coupling routinely mentions bottleneck effects that arise from the narrow windows
of opportunities that wave-cycles offer, or the limited patterns of oscillations
that local structures can sustain. In independent work we examine how the
frequency-couplings put forward here may capture some of the basic restrictions found cross-linguistically, with special attention to ‘local’ patterns of
exclusion known as “anti-locality” or “identity avoidance” (Richards, 2010,
Boeckx, 2014).
As already pointed out above, the neurobiological architecture proposed
here departs from traditional thinking about language and its implementation in the brain. But we wish to stress that we regard some of the betterestablished claims in neurolinguistics to complement our approach. For instance, although we question the claim that “syntax” crucially relies on Broca’s
region (Friederici, 2009), we recognize that Broca’s region plays a crucial role in
processing hierarchical representations (specifically, hierarchical sequences, in
the sense of Fitch and Martins (2014)), as shown in Pallier et al. (2011), Ohta
et al. (2013). We take Broca’s area to provide a memory stack that is independently needed to linearize complex syntactic structures (Boeckx et al., 2014)
and to integrate or unify several linguistic representations (Hagoort, 2005,
Flinker et al., 2015), including sound and meaning (Bornkessel-Schlesewsky
and Schlesewsky, 2013).
Likewise, although we do not endorse the view that the anterior temporal
lobe is the locus of syntactic structure building (Brennan et al., 2012), we endorse the more nuanced conclusion that this region plays a crucial role in compositional interpretation of syntactic structures (Westerlund and Pylkk¨anen,
2014, Del Prato and Pylkk¨
anen, 2014). Indeed, our conception of PSBC leads
us to take the canonical fronto-temporal language network to be an output
system of the combinatorial processes we have focused on here. We think that
it is likely that this network, rooted as it appears to be in primate audition
(Bornkessel-Schlesewsky et al., 2015), was recruited (in the sense of Dehaene
(2005), Parkinson and Wheatley (2013)) to pair sound and meaning boosted
up by PSBC in our species. It also bears repeating that in line with current
linguistic theorizing we take the PSBC to be dissociable from phonological
processing. As a result, the involvement of some of the frequency ranges we
have appealed to here may not have been detected in standard neurolinguistic
paradigms, which adopt too concrete a view on linguistic structures, too close
to auditory processing. Such a reliance on concrete perceptual events may not
provide the right detection threshholds for pure syntactic computations.
Quite apart from this complementarity, we think that a major conclusion
of our approach lies in the role of subcortical structures in language beyond the
domain of speech (Kotz and Schwartze, 2010). We anticipate that properties
once thought to be the exclusivity of the cortex will turn out to be mirrored
sub cortically, as recent research indicates (Friederici, 2006, Wahl et al., 2008,
Jeon et al., 2014).
Antzoulatos, Evan G, and Earl K Miller. 2014. Increases in functional connectivity between prefrontal cortex and striatum during category learning.
Neuron 83:216–225.
Bastos, Andre M, W Martin Usrey, Rick A Adams, George R Mangun, Pascal Fries, and Karl J Friston. 2012. Canonical microcircuits for predictive
coding. Neuron 76:695–711.
Bastos, Andre M, Julien Vezoli, and Pascal Fries. 2015. Communication
through coherence with inter-areal delays. Current opinion in neurobiology 31:173–180.
Ben Shalom, Dorit, and David Poeppel. 2008. Functional anatomic models of
language: assembling the pieces. The Neuroscientist 14:119–127.
Boeckx, Cedric. 2011. Some reflections on Darwin’s Problem in the context
of Cartesian Biolinguistics. In The biolinguistic enterprise: New perspectives on the evolution and nature of the human language faculty, ed. A.-M.
Di Sciullo and C. Boeckx, 42–64. Oxford: Oxford University Press.
Boeckx, Cedric. 2014. Elementary syntactic structures. Cambridge: Cambridge University Press.
Boeckx, Cedric, and Antonio Ben´ıtez-Burraco. 2014. The shape of the
language-ready brain. Frontiers in Psychology 5:282.
Boeckx, Cedric, Anna Martinez-Alvarez, and Evelina Leivada. 2014. The functional neuroanatomy of serial order in language. Journal of Neurolinguistics
Bollimunta, Anil, Jue Mo, Charles E Schroeder, and Mingzhou Ding. 2011.
Neuronal mechanisms and attentional modulation of corticothalamic alpha
oscillations. The Journal of Neuroscience 31:4935–4943.
Borer, Hagit. 2005. Structuring Sense (2 vols.). Oxford: Oxford University
Bornkessel-Schlesewsky, Ina, and Matthias Schlesewsky. 2013. Reconciling
time, space and function: a new dorsal–ventral stream model of sentence
comprehension. Brain and language 125:60–76.
Bornkessel-Schlesewsky, Ina, Matthias Schlesewsky, Steven L Small, and
Josef P Rauschecker. 2015. Neurobiological roots of language in primate
audition: common computational properties. Trends in Cognitive Sciences
Bosman, Conrado A, Carien S Lansink, and Cyriel Pennartz. 2014. Functions
of gamma-band synchronization in cognition: from single circuits to functional diversity across cortical and subcortical systems. European Journal
of Neuroscience 39:1982–1999.
Brennan, Jonathan, Yuval Nir, Uri Hasson, Rafael Malach, David J Heeger,
and Liina Pylkk¨
anen. 2012. Syntactic structure building in the anterior
temporal lobe during natural story listening. Brain and language 120:163–
Buckner, Randy L, Jessica R Andrews-Hanna, and Daniel L Schacter. 2008.
The brain’s default network. Annals of the New York Academy of Sciences
aki, Gy¨
orgy. 2006. Rhythms of the brain. Oxford: Oxford University
aki, Gy¨
orgy. 2010. Neural syntax: cell assemblies, synapsembles, and
readers. Neuron 68:362–385.
aki, Gy¨
orgy, Nikos Logothetis, and Wolf Singer. 2013. Scaling brain
size, keeping timing: Evolutionary preservation of brain rhythms. Neuron
aki, Gy¨
orgy, and Brendon O Watson. 2012. Brain rhythms and neural
syntax: implications for efficient coding of cognitive content and neuropsychiatric disease. Dialogues in clinical neuroscience 14:345.
Carruthers, P. 2006. The Architecture of the Mind . Oxford: Oxford University
Chan, Jason L, Michael J Koval, Thilo Womelsdorf, Stephen G Lomber, and
Stefan Everling. 2014. Dorsolateral prefrontal cortex deactivation in monkeys reduces preparatory beta and gamma power in the superior colliculus.
Cerebral Cortex bhu154.
Chomsky, Noam. 1957. Syntactic structures. The Hague: Mouton.
Chomsky, Noam. 1970. Remarks on nominalization. In Readings in English transformational grammar , ed. R. Jacobs and P. Rosenbaum, 184–221.
Waltham, Mass.: Ginn and Co.
Chomsky, Noam. 2007. Approaching UG from below. In Interfaces + recursion = language? Chomsky’s minimalism and the view from semantics, ed.
U. Sauerland and H.-M. G¨artner, 1–30. Mouton de Gruyter.
DeCoteau, William E, Catherine Thorn, Daniel J Gibson, Richard Courtemanche, Partha Mitra, Yasuo Kubota, and Ann M Graybiel. 2007. Learningrelated coordination of striatal and hippocampal theta rhythms during acquisition of a procedural maze task. Proceedings of the National Academy
of Sciences 104:5644–5649.
Dehaene, Stanislas. 2005. Evolution of human cortical circuits for reading and
arithmetic: The “neuronal recycling” hypothesis. In From monkey brain to
human brain, ed. S. Dehaene, J.-R. Duhamel, M. Hauser, and J. Rizzolatti,
133–157. Cambridge, Mass.: MIT Press.
Dehaene, Stanislas, Michel Kerszberg, and Jean-Pierre Changeux. 1998. A
neuronal model of a global workspace in effortful cognitive tasks. Proceedings
of the National Academy of Sciences 95:14529–14534.
Deiber, Marie-Pierre, Pascal Missonnier, Olivier Bertrand, Gabriel Gold, Lara
Fazio-Costa, Vicente Ibanez, and Panteleimon Giannakopoulos. 2007. Distinction between perceptual and attentional processing in working memory
tasks: a study of phase-locked and induced oscillatory brain dynamics. Journal of cognitive neuroscience 19:158–172.
Del Prato, Paul, and Liina Pylkk¨anen. 2014. Meg evidence for conceptual
combination but not numeral quantification in the left anterior temporal
lobe during language production. Frontiers in psychology 5.
Dipoppa, Mario, and Boris S Gutkin. 2013. Flexible frequency control of cortical oscillations enables computations required for working memory. Proceedings of the National Academy of Sciences 110:12828–12833.
Dugu´e, Laura, Philippe Marque, and Rufin VanRullen. 2014. Theta oscillations
modulate attentional search performance periodically. Journal of Cognitive
Neuroscience 26:1–14.
Duncan, John. 2010. The multiple-demand (md) system of the primate brain:
mental programs for intelligent behaviour. Trends in cognitive sciences
Duncan, John. 2013. The structure of cognition: attentional episodes in mind
and brain. Neuron 80:35–50.
Embick, David, and David Poeppel. 2014. Towards a computational(ist) neurobiology of language: correlational, integrated and explanatory neurolinguistics. Language, Cognition and Neuroscience 1–10.
Engel, Andreas K, and Pascal Fries. 2010. Beta-band oscillations—signalling
the status quo? Current opinion in neurobiology 20:156–165.
Fedorenko, E., and S. L. Thompson-Schill. 2014. Reworking the language
network. Trends in cognitive sciences 18:120–126.
Fitch, W. Tecumseh, and Mauricio D Martins. 2014. Hierarchical processing
in music, language, and action: Lashley revisited. Annals of the New York
Academy of Sciences 1316:87–104.
Flinker, Adeen, Anna Korzeniewska, Avgusta Y Shestyuk, Piotr J
Franaszczuk, Nina F Dronkers, Robert T Knight, and Nathan E Crone.
2015. Redefining the role of broca’s area in speech. Proceedings of the
National Academy of Sciences 112:2871–2875.
Fox, Michael D, Abraham Z Snyder, Justin L Vincent, Maurizio Corbetta,
David C Van Essen, and Marcus E Raichle. 2005. The human brain is intrinsically organized into dynamic, anticorrelated functional networks. Proceedings of the National Academy of Sciences of the United States of America
Friederici, Angela D. 2006. What’s in control of language? Nature neuroscience
Friederici, Angela D. 2009. Brain circuits of syntax. In Biological Foundations
and Origin of Syntax. Str¨
ungmann Forum Reports (Volume 3), 239–252.
Mit Press.
Friederici, Angela D, and Sarah ME Gierhan. 2013. The language network.
Current Opinion in Neurobiology 23:250–254.
Giraud, Anne-Lise, and David Poeppel. 2012. Cortical oscillations and speech
processing: emerging computational principles and operations. Nature neuroscience 15:511–517.
Hagoort, Peter. 2005. On broca, brain, and binding: a new framework. Trends
in cognitive sciences 9:416–423.
Hagoort, Peter. 2014. Nodes and networks in the neural architecture for language: Broca’s region and beyond. Current opinion in neurobiology 28:136–
Hale, Ken, and S. Jay Keyser. 1993. On argument structure and the lexical
expression of grammatical relations. In The view from Building 20: Essays
in Linguistics in honor of Sylvain Bromberger , ed. K. Hale and S. J. Keyser,
53–110. Cambridge, Mass.: MIT Press.
Halle, Morris, and Alec Marantz. 1993. Distributed morphology and the pieces
of inflection. In The view from building 20 , ed. K. Hale and S. J. Keyser,
111–176. Cambridge, Mass.: MIT Press.
Hauser, Marc D. 2009. The possibility of impossible cultures. Nature 460:190–
Honkanen, Roosa, Santeri Rouhinen, Sheng H Wang, J Matias Palva, and
Satu Palva. 2014. Gamma oscillations underlie the maintenance of featurespecific information and the contents of visual working memory. Cerebral
Cortex bhu263.
Huguenard, John R, and David A McCormick. 2007. Thalamic synchrony and
dynamic regulation of global forebrain oscillations. Trends in neurosciences
Jeon, Hyeon-Ae, Alfred Anwander, and Angela D Friederici. 2014. Functional
network mirrored in the prefrontal cortex, caudate nucleus, and thalamus:
High-resolution functional imaging and structural connectivity. The Journal
of Neuroscience 34:9202–9212.
Kayne, Richard S. 1984. Connectedness and binary branching. Dordrecht:
Kayne, Richard S. 1994. The antisymmetry of syntax . Cambridge, Mass.:
MIT Press.
Klimesch, Wolfgang. 2012. Alpha-band oscillations, attention, and controlled
access to stored information. Trends in cognitive sciences 16:606–617.
Kotz, Sonja A, and Michael Schwartze. 2010. Cortical speech processing unplugged: a timely subcortico-cortical framework. Trends in cognitive sciences 14:392–399.
Leventhal, Daniel K, Gregory J Gage, Robert Schmidt, Jeffrey R Pettibone,
Alaina C Case, and Joshua D Berke. 2012. Basal ganglia beta oscillations
accompany cue utilization. Neuron 73:523–536.
Lisman, John. 2005. The theta/gamma discrete phase code occuring during the
hippocampal phase precession may be a more general brain coding scheme.
Hippocampus 15:913–922.
Marantz, Alec. 2008. Words and phases. In Phases in the theory of grammar ,
ed. S.-H. Choe, 191–222. Seoul: Dong In.
Martin, Claire, and Nadine Ravel. 2014. Beta and gamma oscillatory activities associated with olfactory memory tasks: different rhythms for different
functional networks? Frontiers in behavioral neuroscience 8.
Miller, Earl K., and Tim J. Buschman. 2012. Cortical circuits for the control
of attention. Current opinion in neurobiology 23:1–7.
Mithen, Steven J. 1996. The prehistory of the mind . London: Thames and
Ohta, Shinri, Naoki Fukui, and Kuniyoshi Sakai. 2013. Syntactic computation
in the human brain: The degree of merger as a key factor. PloS one 8:e56230.
Pallier, Christophe, Anne-Dominique Devauchelle, and Stanislas Dehaene.
2011. Cortical representation of the constituent structure of sentences. Proceedings of the National Academy of Sciences 108:2522–2527.
Palva, Satu, and J Matias Palva. 2011. Functional roles of alpha-band phase
synchronization in local and large-scale cortical networks. Frontiers in psychology 2.
Parkinson, Carolyn, and Thalia Wheatley. 2013. Old cortex, new contexts:
re-purposing spatial perception for social cognition. Frontiers in human
neuroscience 7.
Parnaudeau, Sebastien, Pia-Kelsey O’Neill, Scott S Bolkan, Ryan D Ward,
Atheir I Abbas, Bryan L Roth, Peter D Balsam, Joshua A Gordon, and
Christoph Kellendonk. 2013. Inhibition of mediodorsal thalamus disrupts
thalamofrontal connectivity and cognition. Neuron 77:1151–1162.
de Pasquale, Francesco, Stefania Della Penna, Abraham Z Snyder, Laura
Marzetti, Vittorio Pizzella, Gian Luca Romani, and Maurizio Corbetta.
2012. A cortical core for dynamic integration of functional networks in the
resting human brain. Neuron 74:753–764.
Pearce, Eiluned, Chris Stringer, and Robin I. M. Dunbar. 2013. New insights
into differences in brain organization between neanderthals and anatomically
modern humans. Proceedings of the Royal Society of London. Series B,
Biological Sciences 280:20130168.
Pietroski, Paul M. 2012. Language and conceptual reanalysis. In Towards a
Biolinguistic Understanding of Grammar: Essays on Interfaces, ed. A.M.
Di Sciullo. John Benjamins.
Poeppel, David. 2014. The neuroanatomic and neurophysiological infrastructure for speech and language. Current opinion in neurobiology 28:142–149.
Poeppel, David, and David Embick. 2005. Defining the relation between linguistics and neuroscience. In Twenty-first century psycholinguistics: Four
cornerstones, ed. A. Cutler, 173–189. Hillsdale, NJ:: Erlbaum.
Poeppel, David, Karen Emmorey, Gregory Hickok, and Liina Pylkk¨anen. 2012.
Towards a new neurobiology of language. The Journal of Neuroscience
Raichle, Marcus E, Ann Mary MacLeod, Abraham Z Snyder, William J Powers, Debra A Gusnard, and Gordon L Shulman. 2001. A default mode of
brain function. Proceedings of the National Academy of Sciences 98:676–682.
Ravignani, Andrea, Daniel L. Bowling, and W. Tecumseh Fitch. 2014. Chorusing, synchrony, and the evolutionary functions of rhythm. Frontiers in
psychology 5.
Reinhart, Tanya. 2006. Interface strategies: optimal and costly computations.
Cambridge, Mass.: MIT Press.
Richards, Norvin. 2010. Uttering trees. Cambridge, Mass.: MIT Press.
Roux, Fr´ed´eric, and Peter J Uhlhaas. 2014. Working memory and neural
oscillations: alpha–gamma versus theta–gamma codes for distinct wm information? Trends in cognitive sciences 18:16–25.
Roux, Fr´ed´eric, Michael Wibral, Wolf Singer, Jaan Aru, and Peter J Uhlhaas.
2013. The phase of thalamic alpha activity modulates cortical gammaband activity: evidence from resting-state meg recordings. The Journal of
Neuroscience 33:17827–17835.
Rubin, Rachael D, Patrick D Watson, Melissa C Duff, and Neal J Cohen.
2014. The role of the hippocampus in flexible cognition and social behavior.
Frontiers in human neuroscience 8.
Saalmann, Yuri B, and Sabine Kastner. 2011. Cognitive and perceptual functions of the visual thalamus. Neuron 71:209–223.
Saalmann, Yuri B, Mark A Pinsk, Liang Wang, Xin Li, and Sabine Kastner. 2012. The pulvinar regulates information transmission between cortical
areas based on attention demands. Science 337:753–756.
Salami, Mahmoud, Chiaki Itami, Tadaharu Tsumoto, and Fumitaka Kimura.
2003. Change of conduction velocity by regional myelination yields constant
latency irrespective of distance between thalamus and cortex. Proceedings
of the National Academy of Sciences 100:6174–6179.
Salazar, RF, NM Dotson, SL Bressler, and CM Gray. 2012. Content-specific
fronto-parietal synchronization during visual working memory. Science
Schaafsma, Sara M., Donald W. Pfaff, Robert P. Spunt, and Ralph Adolphs.
2015. Deconstructing and reconstructing theory of mind. Trends in cognitive
sciences 19:65–72.
Shanahan, M. 2012. The brain’s connective core and its role in animal cognition. Philosophical Transactions of the Royal Society B: Biological Sciences
Song, Kun, Ming Meng, Lin Chen, Ke Zhou, and Huan Luo. 2014. Behavioral
oscillations in attention: rhythmic α pulses mediated through θ band. The
Journal of Neuroscience 34:4837–4844.
Spelke, Elizabeth. 2003. What makes us smart? Core knowledge and natural
language. In Language and Mind: Advances in the study of language and
thought, ed. D. Gentner and S. Goldin-Meadow, 277–311. Cambridge, Mass.:
MIT Press.
Spelke, Elizabeth, and Katherine D. Kinzler. 2007. Core knowledge. Developmental science 10:89–96.
Tallon-Baudry, Catherine, Sunita Mandon, Winrich A Freiwald, and Andreas K Kreiter. 2004. Oscillatory synchrony in the monkey temporal lobe
correlates with performance in a visual short-term memory task. Cerebral
Cortex 14:713–720.
Theofanopoulou, Constantina, and Cedric Boeckx. To appear. The central role
of the thalamus in language and cognition. In Advances in biolinguistics,
ed. C. Boeckx and K. Fujita. London: Routledge.
Theyel, Brian B, Daniel A Llano, and S Murray Sherman. 2010. The corticothalamocortical circuit drives higher-order cortex in the mouse. Nature
neuroscience 13:84–88.
Tort, Adriano BL, Mark A Kramer, Catherine Thorn, Daniel J Gibson, Yasuo Kubota, Ann M Graybiel, and Nancy J Kopell. 2008. Dynamic cross-
frequency couplings of local field potential oscillations in rat striatum and
hippocampus during performance of a T-maze task. Proceedings of the National Academy of Sciences 105:20517–20522.
Vertes, Robert P, Walter B Hoover, and Gonzalo Viana Di Prisco. 2004. Theta
rhythm of the hippocampus: subcortical control and functional significance.
Behavioral and cognitive neuroscience reviews 3:173–200.
Vicente, Raul, Leonardo L Gollo, Claudio R Mirasso, Ingo Fischer, and Gordon
Pipa. 2008. Dynamical relaying can yield zero time lag neuronal synchrony
despite long conduction delays. Proceedings of the National Academy of
Sciences 105:17157–17162.
Von Stein, Astrid, and Johannes Sarnthein. 2000. Different frequencies for
different scales of cortical integration: from local gamma to long range
alpha/theta synchronization. International Journal of Psychophysiology
Wahl, Michael, Frank Marzinzik, Angela D Friederici, Anja Hahne, Andreas
Kupsch, Gerd-Helge Schneider, Douglas Saddy, Gabriel Curio, and Fabian
Klostermann. 2008. The human thalamus processes syntactic and semantic
language violations. Neuron 59:695–707.
Westerlund, Masha, and Liina Pylkk¨anen. 2014. The role of the left anterior
temporal lobe in semantic composition vs. semantic memory. Neuropsychologia 57:59–70.
Womelsdorf, Thilo, Salva Ardid, Stefan Everling, and Taufik A Valiante. 2014.
Burst firing synchronizes prefrontal and anterior cingulate cortex during
attentional control. Current Biology 24:2613–2621.
Figure 1: Areas, connections, and rhythm generations
Figure 2: Key to Figure 1
Figure 3: Cross-frequency couplings mapping onto Phrase Structural relations