How many brain regions are needed to elucidate the

neural bases of fear and anxiety?

Luiz Pessoa
Department of Psychology, Department of Electrical and Computer Engineering, Maryland Neuroimaging
Center, University of Maryland, College Park, MD 20742, USA
pessoa@umd.edu

Abstract
We suggest that to understand complex behaviors associated with fear and anxiety, we need to understand
brain processes at the collective, network level. But what should be the type and spatial scale of the targeted
circuits/networks? Not only are multi-region interactions essential— including complex reciprocal interactions,
loops, and other types of arrangement— but it is profitable to characterize circuits spanning the entire
neuroaxis. In particular, it is productive to conceptualize the circuits contributing to fear/anxiety as embedded
into large-scale connectional systems. We discuss circuits involving the basolateral amygdala that contribute
to aversive conditioning and fear extinction. In addition, we highlight the importance of the extended
amygdala (central nucleus of the amygdala and bed nucleus of the stria terminalis) cortical-subcortical loop,
which allows large swaths of cortex and subcortex to influence fear and anxiety. In this manner, fear/anxiety
canbeunderstoodnotonlybasedontraditional“descending”mechanismsinvolvingthehypoth
brainstem, but in terms of a considerably broader reentrant organization.

Keywords: fear, anxiety, extended amygdala, bed nucleus of the stria terminalis, networks

1. Introduction
Neuroscientists seek to understand the neural basis of mental functions. What will it take to understand the
brain basis of fear and anxiety? Clearly, understanding one ortwobrainregionswon’tbe—enough we need
to study these constructs at the circuit or network level (we’llusetheterms“circuit”and“network”moreor
less interchangeably, and in a functional sense, not anatomically).Butevenhere,it’snot what spatial
clear
scales are most profitable. And if the answer is a circuit or network, what kind of circuit should be
considered? And what does it mean to study fear and anxiety at the network level?
Before proceeding, it is worth saying a few words about the terms fear”
“ and “anxiety”. In fear, typically,
the danger is imminent, mostly unambiguous, such that the animal is mobilized for immediate action,
including fight or flight. In anxiety, typically, threatsaremoreuncertainanddiffuse.It’s izedoften
that emphas
anxiety involves a lasting state of apprehension of potential future threats, accompanied by negative affect,
autonomic symptoms, worry, increased vigilance, and passive avoidance.
However, the above conceptualization, common as it is, is problematic because it encourages a fairly
binary division of labor. As developed elsewhere (Pessoa et al., 2022), whereas mental terms can be at times
useful in orienting researchers along research avenues, they are generally inadequate in conveying the
interdependence of mental processes. The discussion of neural circuits below should help illustrate how
neuralcircuitsdonotrespectboundariestypicallyadoptedbyinvestigators.Accordingly,“fea

1
are used as placeholders but should not be understood as dichotomous constructs that map to separate
neural circuits.
In the present piece, most of the literature on the neural basis of fear and anxiety is described based on
the rodent literature, which provides a more comprehensive picture of the circuits involved. Human studies
paint a somewhat similar picture, but space limitations preclude a detailed interspecies comparison.
Nevertheless, the ensuing discussion applies across species.

2. Circuits involved in fear-related processing

For brevity, we will illustrate fear-related processing with classical conditioning circuits (LeDoux, 2000; Maren,
2001; Tovote et al., 2015). Asstatedabove,wedonotmeantoimplyastrictseparationof“fear”and
“anxiety”,whichwebelieveiscounter productive. This should not be surprising, after all brain circuits have
been shaped by evolution to solve behavioral problems that promote survival. Mental terms used by human
researchers are, consequently, poor descriptors that have more to do with research traditions than anything
else.
A critical node of classical conditioning circuits is the amygdala, a highly heterogenous complex with a
dozen or more anatomical subdivisions. Here, we will focus on the basolateral amygdala and the central
nucleus of the amygdala. The central nucleus targets multiple areas along the basal forebrain and brainstem
that mobilize the neuroendocrine and autonomic systems. How does the central nucleus itself get engaged?
During conditioning, the lateral component of the basolateral amygdala is essential for learning the
association between the conditioned stimulus (CS+; which initially is benign) and the unconditioned stimulus
(US; inherently aversive). Once learning solidifies, the basolateral amygdala engages the central nucleus
when a CS+ is encountered (Figure 1A). The circuit described thus far, although building upon local
amygdala micro-circuits, can be considered essentially unidirectional: sensory signals linked to the CS+
impinge on the basolateral amygdala, which engages the central amygdala, with outflow to the hypothalamus
and multiple sites along the brainstem. Indeed, this descending organization of fear (and anxiety) circuits is a
hallmark of existing proposals.
Although the basic circuit described elucidates key
components of aversive conditioning, the circuit needs to be
extended (Figure 1B). For example, conditioning based on
more complex auditory stimuli is abolished when auditory
cortex circuits are blocked (Letzkus et al., 2011). Even in
response to simple auditory conditioned stimuli, fear expression
depends on the prelimbic cortex in medial prefrontal cortex
(PFC; dorsomedial PFC in humans) (Corcoran and Quirk,
2007). Furthermore, the prelimbic cortex and the basal
amygdala are reciprocally connected and exhibit entrainment of
theta rhythms after conditioning (Likhtik et al., 2014).
Bidirectional interactions also exist between the basolateral Figure 1. Fear circuits. (A) Traditional
amygdala and the hippocampus, a region that processes circuit focusing on the descending
engagement of autonomic and
context-related information. This is important because fear neuroendocrine responses. (B) Expanded
responses can be context dependent (see below on extinction). circuit with bidirectional connections.
Finally, the paraventricular nucleus of the thalamus (PVT) is Abbreviations: BLA basolateral amygdala;
required for the expression of remote but not recent fear CE, central nucleus of the amygdala;
HIPP, hippocampus; MPFC, medial
memories (Do-Monte et al., 2015). Because the last three prefrontal cortex; PVT, paraventricular
regions discussed— prelimbic cortex, hippocampus, and PVT— nucleus of the thalamus.
are bidirectionally connected with the basolateral amygdala,

2
the coordinated activity between all of them likely plays a notable role in fear-related processing.
The discussion thus far illustrates the need to consider a broader set of brain regions in studying fear-
related processing. (For brevity, we have omitted additional regions, including the periaqueductal gray, which
conveys signals about the US to the central amygdala; Johansen et al., 2010). But it is necessary to go
beyond the“minimalcircuit” in Figure 1B because fear needs to be understood both in terms of processing
that promotes fear and processing that opposes it. Regarding the latter, for example, fear extinction
processes eventually transform a fear-inducing CS+ stimulus into one that is (mostly) neutral (Dunsmoor et
al., 2015; Bouton et al., 2021). Thus, in all but the simplest laboratory settings, fear-promoting and -opposing
processes are at play and need to be considered to explain behavior.
Let us briefly consider extinction. Both behaviorally and neurobiologically, fear extinction is now
understood to be rather complex (Dunsmoor et al., 2015; Bouton et al., 2021). Early models of extinction
emphasized the role of the medial PFC (infralimbic cortex in rodents and ventromedial PFC in humans) in
modulating the basolateral amygdala to dampen fear in the face of a now-extinguished stimulus— a previous
CS+ that, through extinction learning, now is associated with safety (Figure 2A). As the safety of a previous
CS+ critically depends on environmental context, the hippocampus was viewed as important, too. Such early
models have been substantially updated, and the emerging picture is considerably more elaborate (Figure
2B). A key development has been the realization that the medial PFC works in a coordinated fashion with the
basolateral amygdala (the two are bidirectionally connected), such that the initial idea that the former
(associated with “cognition” in cortex) simply dampens the latter (associated with “emotion”insubcortex)is
problematic (Figure 2A). Some studies
even suggest that the basolateral
amygdala is“upstream”ofthemedialPFC,
asapopulationof“extinctionneurons”in
the basolateral amygdala increase their
activity during extinction learning (Herry et
al., 2008), and contribute to medial PFC
activity.
To reiterate, the elucidation of the
neural basis of fear requires working out Figure 2. Fear extinction circuits. (A) Basic circuit focusing on
the promotion of fear as well as the regulation of the amygdala by the medial prefrontal cortex. (B)
Expanded circuit with bidirectional connections. Reuniens is a
dampening of fear. The circuits are often nucleus of the thalamus. Arrow in blue represent indirect
studied separately, but should be connections. Abbreviations: PAG, periaqueductal gray; VTA,
considered jointly for a comprehensive ventral tegmental area. See also Figure 1.
view of fear processing.

3. Circuits involved in anxiety-related processing

Circuits involved in anxiety-related processing overlap with those in fear but differences must be highlighted.
One of the most noteworthy concerns the bed nucleus of the stria terminalis (BST; also abbreviated as
BNST). In an influential model, the BST was proposed to be engaged by diffuse threat, whereas the central
amygdala was proposed to be engaged by immediate threat (Davis and Whalen, 2001). This purported
dissociation has been challenged by some researchers (Gungor and Pare, 2016; Shackman and Fox, 2016;
Fox and Shackman, 2019), but the precise contributions of the BST and the central amygdala to sustained
versus phasic processes, respectively, remains unresolved (for examples in humans, see Hur et al., 2020;
Murty et al., 2022).
Another brain region central to anxiety-related processes is the insula. Indeed, in the human literature,
models of anxiety often concur in emphasizing the key role of this region, especially the anterior sector
(Paulus and Stein, 2007; Grupe and Nitschke, 2013). Intriguingly, frameworks centered on rodents

3
emphasize subcortical contributions (with the clear exception of the medial PFC), such that the insula in many
cases is entirely missing, although recent studies have started to investigate the contributions of this region in
rodents, too (Klein et al., 2021). At present, the roles of the insula in anxiety remain somewhat unclear, but
have been proposed to include heightened responses during the anticipation of aversive events and the
evaluation of risk. In addition, the insula is believed to generate anticipatory responses in the face of
hypothetical future events so as to answer the following question: “howisitgoingtofeel?”(Grupe and
Nitschke, 2013).
From a broader perspective, uncovering the neural basis of anxiety poses multiple challenges. To see
why, consider multiple ways in which threat-related processing is believed to contribute to maladaptive
behaviors in humans with anxiety disorders (Grupe and Nitschke, 2013): inflated estimates of threat
magnitude and probability, hypervigilance, deficient safety learning, behavioral and cognitive avoidance, and
heightened reactivity to threat uncertainty. The broad range of these processes demonstrates the
multifaceted nature of “anxiety” — ultimately, a broad umbrella term. Consequently, we can say that there is
no“anxietynetwork” (singular), as much as a variety of circuits that contribute to anxiety-related behavioral
manifestations.

4. Embedding circuits into large-scale systems

The multifaceted aspects of fear and anxiety motivate understanding them in an even broader anatomical-
functional perspective. Accordingly, it is profitable to situate them in terms of the connectional logic of the
neuroarchitecture that interlinks the multiple sectors of the neuroaxis (Pessoa et al., 2019, discusses how this
organization is present across most vertebrates). A key element of this organization is the presence of
cortical-subcortical-cortical loops, as exemplified by basal ganglia loops.
Basal ganglia loops are a defining feature of the architecture of the vertebrate brain (except in fishes).
In mammals, the entire cortical sheet projects to the striatum and loops back to the cortex via the thalamus
(Figure 3A). An important feature of mammalian basal ganglia loops is that they involve both dorsal (caudate-
putamen) and ventral (nucleus accumbens) striatal components. Some cortical areas project to the dorsal
striatum (for example, motor and somatosensory areas), while others project to the ventral striatum (in
primates, for example, orbitofrontal, prefrontal, and anterior cingulate cortices).
Notably, a similar connectional logic is observed involving other parts of the basal forebrain, of which
we emphasize cortical-subcortical
loops engaging the“extended
amygdala” . The extended amygdala
concept developed by Alheid and
Heimer (1988) considers the central
amygdala and the BST to be a basal-
ganglia-like anatomical-functional
unit. To appreciate this organization,
it is important to consider the two
major components of the forebrain: Figure 3. Cortical-subcortical loops are an important principle of
the pallium and the subpallium. macro-scale anatomical organization. (A) Standard basal ganglia
During embryological development, loops. All sectors of the cortex project to the striatum, looping bask via
the pallium gives rise not only to the the thalamus. (B) Extended amygdala loop has the same overall
entire cortex but also to the organization. Note that the most substantial projection from the cortex/
pallium is from the basolateral amygdala which is substantially more
basolateral amygdala, whereas the
pronounced than that of other sectors. Line thickness of the
subpallium gives rise to the connections between the cortex/pallium to the central amygdala
subcortex, including the central conveys pathway weight.
amygdala and the BST. Thus, we

4
see that the latter two regions belong to a qualitatively different sector of the brain compared to the
basolateral amygdala (and cortex). In addition, based on cell types and molecular profiles, the central
amygdala is a striatum-like region, whereas the BST is a pallidum-like region (in mammals, the pallidum
corresponds to the globus pallidus).
With the above facts in mind, now it should be possible to follow the extended amygdala loop (Figure
3B). The basolateral (pallial) amygdala interfaces with the extended amygdala much like the cortex interfaces
with standard basal ganglia loops (functionally, this also matches the integrative properties of the pallial
amygdala, which receives massive inputs from across the cortex; see below). The central amygdala
(striatum-like region) projects to the BST (pallidum-like region). The BST subsequently projects to the
thalamus, which in turn projects to cor tical targets. The pathways from the BST to the thalamus target the
PVT and other midline nuclei. In all, the overall arrangement establishes a pathway through the central
extended amygdala and back to the cortex/pallium. For a detailed exposition of the extended amygdala
system (Figure 3B), see Heimer et al. (2007).
Standard basal ganglia loops (via the striatum) play a major role in the flow of cortical signaling.
Classically linked to movement control and disorders, the basal ganglia are now known to be involved in
cognition, motivation, and emotion, and viewed as essential for higher level behavioral control, including
learning and regulation of stimulus-driven behaviors, as well as action selection supporting goal-directed
behaviors (Yin and Knowlton, 2006; DeLong and Wichmann, 2009; Nelson and Kreitzer, 2014). We propose
that the extended amygdala loop should be conceptualized as contributing to a broad and diverse set of
cognitive-emotional-motivational processes, too.
What is the importance of the extended amygdala loop in the case of fear and anxiety? Traditionally,
conceptualizations of fear and anxiety circuits center around two key properties. First, they are highly
centralized. For example, fear circuits are centered on the basolateral and central amygdala, and anxiety
circuits are centered on the BST. Second, they are built around the idea of descending control (Figure 4A).
For example, both the central amygdala and the BST assemble autonomic and endocrine responses by
engaging the hypothalamus and brainstem (historically, the hypothalamus itself has been conceptualized as a
“mastercontroller” autonomic
ofthesystem). The reentrant organization of the extended amygdala loop
suggests a complementary view that places fear- and anxiety-related processing as embedded within
broader cognitive-emotional-motivational circuits (Figure 4B).
To further motivate this idea, consider the relationship between cortical-subcortical reentrant systems.
Classical basal ganglia loops (via the striatum) are viewed as rather independent and parallel. However,
accruing evidence points to considerable crosstalk between these systems, with substantial signal intermixing
(Joel and Weiner, 1994; Haber, 2010; Hintiryan et al., 2016; Groenewegen et al., 2016; Aoki et al., 2019).
We thus propose that they be viewed as distributed, intercommunicating systems that provide integral
contributions to cortical function. How
about the extended amygdala loop? The
cortical-like component of the loop is the
basolateral amygdala, which is
bidirectionally connected with essentially
the entirety of the cortex (albeit with
different connectivity strengths). This
organization shows that the extended
amygdala loop has unique potential to
contribute to overall brain function.
As stated, classical basal ganglia Figure 4. Contrasting orgnaizations. (A) Tradiontal view in terms
loops are not independent and have of centralized processing and descending control. (B)
multiple opportunities to communicate Complementary proposal in which the reentrant organization of
with one another. Remarkably, these the extended amygdala loop plays a key role.

5
loops are interlinked with the extended amygdala loop, too. We suggest this is a significant arrangement
because it allows the circulation of disparate signals (action-related, emotional, motivational, and so on)
across multiple loops, considerably broadening the range of signal distribution and integration. In particular,
the PVT is well-positioned to interlink systems (Kirouac, 2015) (Figure 5). This thalamic nucleus projects to
both the central extended amygdala and the nucleus accumbens, and is reciprocally connected with pallial
areas, such as the insular cortex, the prefrontal cortex (including orbitofrontal cortex), the hippocampal
formation, and the basolateral complex of the amygdala (these pallial sectors are reciprocally interconnected
and project to the central extended amygdala and nucleus accumbens) (reviewed by Kirouac, 2015, 2021).
Notably, individual PVT neurons have axons that bifurcate to innervate multiple targets (Unzai et al., 2015;
Dong et al., 2017). Most neurons in the PVT project to the nucleus accumbens, but a significant proportion
send collaterals to the BST and the central amygdala. In addition, neurons that project to the nucleus
accumbens, BST, and central amygdala are intermixed throughout the PVT and do not appear to form
clusters of unique subpopulations of projection-specific neurons.
To conclude this section, a brief aside on network science. Hub regions are highly connected ones that
have the potential to play chief roles in signal distribution and integration, with connector hubs having
particular importance in interlinking
disparate parts of the system (Figure 6A).
Here, we propose to extend this notion to
hub circuits, those that have strong
potential in influencing processing in
disparate brain sectors (Figure 6B). We
propose that the extended amygdala loop
is one such hub circuit. Given the
expedient access that the central
amygdala and the BST have to
neuroendocrine and autonomic functions
(via the hypothalamus and brainstem),
the loop places extended amygdala
function in very close association with the Figure 5. Large-scale connectional system intercommunication.
cortex. In this context, attempting to The paraventricular nucleus of the thalamus (PVT) serves as a
hub region that interlinks the central amygdala loop with the
separate emotion, motivation, and
standard basal ganglia ventral loop, both at the level of the
cognition becomes a largely problematic thalamus and cortex/pallium.
exercise.

5. Where next?
The upshot of the ideas developed here is that to understand complex behaviors we associate with fear and
anxiety, we need to understand brain processes at the collective, network level. Multi-region interactions are
essential, including complex reciprocal interactions, loops, and other types of arrangement.
So, what are network properties? When the functional unit of interest is apparent only when one
considers the system, but not its component parts, we can say that we have a network-level property. The
case made in the present piece is that in studying fear and anxiety it will not only be profitable but necessary
to consider multiregion functions. For further discussion, please see Pessoa (2014; 2022).
What are additional implications of the ideas described in the present article? Neuroscience is
experiencing a methodological renaissance. Advances in chemistry and genetics now allow precision in
targeting regions and circuits in ways that would have been impossible a decade ago. Multiple developments
permit recording over a larger number of regions simultaneously. We believe such methods will be essential in
advancing the study of fear and anxiety. We have argued elsewhere that it will be important to also develop

6
techniques that enable multiregion perturbations, including activating and/or silencing multiple regions
simultaneously (Pessoa, 2022). We suggest that perturbation experiments could be used to test the
contributions of the extended amygdala cortical-subcorticalloop.Inparticular,inactivatingthe“return
via the thalamus to the cortex/pallium is anticipated to strongly compromise the function of the circuit given
the broad contributions of the return connections to cortical/pallial signals.
Another recommendation is that the field needs
to adopt more dynamic, richly contextual, and
naturalistic experimental designs. In such settings,
we believe that the functions of large-scale systems
discussed here will prove informative. This is because
thebrain’ s considerable anatomical-functional
interactional complexity parallels the enormous
richness of animal behavior (Pessoa et al., 2022).
Typical laboratory settings have severely limited what
can be studied. For example, a type of behavior that Figure 6. Hubs in the brain. (A) Hub regions are
“fits inside a box” is classical conditioning, which has highly connected. (B) Hub circuits are functional units
that can be engaged by or engage multiple ciruits.
been repeatedly examined since the early twentieth
century. Limited in-a-box behaviors have also been
studie d to inform anxiety-related processes. These experimental manipulations offer a window into a few
dimensions of fear and anxiety while allowing careful control over study variables. But the fixation with simple
tasks has led to a form of tunnel vision. As Dennis Paré and Gregory Quirk, prominent researchers in this
area, state in the context of classical conditioning:
When a rat is presented with only one threatening stimulus in a testing box that allows for a single reflexive behavioral response,
one is bound to find exactly what the experimental situation allows: neuronal responses that appear tightly linked to the CS and
seem to obligatorily elicit the conditioned behavior. (Paré and Quirk 2017, 6)

Placed inside a small, enclosed chamber the animal is limited to a sole response: Upon detecting the CS+, it
ceases all overt behavior and freezes in place (see also Holley and Fox, 2022).Itcan’tconsiderotheroptions,
such as dashing to a corner to escape; it cannot try to attack the source of threat either, as there is no other
animal around— the shock comes out of nowhere! Now, whenresearchersstudytherat’sbrainundersuch
conditions, a close relationship between brain and behavior is established. But as Paré and Quirk warn, the
tight link might be apparent insofar as it would not hold under more general conditions.
Thus, while critical, the use of novel neurotechniques mentioned above is insufficient. If we continue
using the paradigms that have been the mainstay of the field, we will be cornering ourselves into a scientific
cul-de-sac. It is time to think outside the box. Fortunately, more naturalistic paradigms are now possible given
recent technical advances. And if we follow novel research paths, it will become apparent that the question of
how many brain regions are needed to understand the neural basis of fear and anxiety is actually ill-posed.
We need to study a large set of intersecting circuit interactions to make progress.

Acknowledgements
The author acknowledges research support from National Institute of Mental Health (MH071589 and
MH112517). The author is also grateful for the constructive feedback provided by reviewers, which
considerably improved the text.

7
References
Alheid, G. F., & Heimer, L. (1988). New perspectives in basal forebrain organization of special relevance for
neuropsychiatric disorders: the striatopallidal, amygdaloid, and corticopetal components of substantia
innominata. Neuroscience, 27(1), 1-39.

Aoki S, Smith JB, Li H, Yan X, Igarashi M, Coulon P, Wickens JR, Ruigrok TJH, Jin X. 2019 An open cortico-
basal ganglia loop allows limbic control over motor output via the nigrothalamic pathway. eLife 8, e49995.
(doi:10.7554/eLife.49995.001)

Bouton, M. E., Maren, S., & McNally, G. P. (2021). Behavioral and neurobiological mechanisms of Pavlovian
and instrumental extinction learning. Physiological Reviews, 101, 611– 681.

Corcoran, K. A., & Quirk, G. J. (2007). Activity in prelimbic cortex is necessary for the expression of learned,
but not innate, fears. Journal of Neuroscience, 27(4), 840-844.

Davis, M., & Whalen, P. J. (2001). The amygdala: vigilance and emotion. Molecular Psychiatry, 6(1), 13-34.

DeLong, M., & Wichmann, T. (2009). Update on models of basal ganglia function and dysfunction.
Parkinsonism & Related Disorders, 15, S237-S240.

Do-Monte, F. H., Quinones-Laracuente, K., & Quirk, G. J. (2015). A temporal shift in the circuits mediating
retrieval of fear memory. Nature, 519(7544), 460-463

Dong, X., Li, S., and Kirouac, G. J. (2017). Collateralization of projections from the paraventricular nucleus of
the thalamus to the nucleus accumbens, bed nucleus of the stria terminalis, and central nucleus of the
amygdala. Brain Struct. Funct. 222, 3927– 3943. doi: 10.1007/s00429-017-1445-8

Dunsmoor, J. E., Niv, Y., Daw, N., & Phelps, E. A. (2015). Rethinking extinction. Neuron, 88(1), 47-63.

Fox, A. S., & Shackman, A. J. (2019). The central extended amygdala in fear and anxiety: Closing the gap
between mechanistic and neuroimaging research. Neuroscience Letters, 693, 58-67.

Groenewegen HJ, Wouterlood FG, Uylings HBM. 2016 Organization of prefrontal– striatal connections. In
Handbook of basal ganglia structure and function (eds H Steiner, K Tseng), chapter 21, pp. 423– 438, 2nd
edn. Amsterdam, The Netherlands: Elsevier.

Grupe, D. W., & Nitschke, J. B. (2013). Uncertainty and anticipation in anxiety: an integrated neurobiological
and psychological perspective. Nature Reviews Neuroscience, 14(7), 488-501.

Gungor, N. Z., & Paré, D. (2016). Functional heterogeneity in the bed nucleus of the stria terminalis. Journal
of Neuroscience, 36(31), 8038-8049.

8
Joel D, Weiner I. 1994 The organization of the basal ganglia– thalamocortical circuits: open interconnected
rather than closed segregated. Neuroscience 63, 363– 379. (doi:10.1016/0306-4522(94)90536-3)

Johansen, J. P., Tarpley, J. W., LeDoux, J. E., & Blair, H. T. (2010). Neural substrates for expectation-
modulated fear learning in the amygdala and periaqueductal gray. Nature Neuroscience, 13(8), 979-986.

Haber SN. 2010 Integrative networks across basal ganglia circuits. In Handbook of basal ganglia structure
and function (eds H Steiner, K Tseng), chapter 24, pp. 409– 427. Amsterdam, The Netherlands: Elsevier.

Heimer, L., Van Hoesen, G. W., Trimble, M., & Zahm, D. S. (2007). Anatomy of neuropsychiatry: the new
anatomy of the basal forebrain and its implications for neuropsychiatric illness. Academic Press.

Herry, C., Ciocchi, S., Senn, V., Demmou, L., Müller, C., & Lüthi, A. (2008). Switching on and off fear by
distinct neuronal circuits. Nature, 454(7204), 600-606.

Hintiryan, H et al. 2016 The mouse cortico-striatal projectome. Nature Neuroscience, 19, 1100– 1114.
(doi:10.1038/nn.4332)

Holley, D., & Fox, A. S. (2022). The central extended amygdala guides survival-relevant tradeoffs:
Implications for understanding common psychiatric disorders. Neuroscience & Biobehavioral Reviews,
104879.

Hur, J., Smith, J. F., DeYoung, K. A., Anderson, A. S., Kuang, J., Kim, H. C., ... & Shackman, A. J. (2020).
Anxiety and the neurobiology of temporally uncertain threat anticipation. Journal of Neuroscience, 40(41),
7949-7964.

Kirouac, G. J. (2015). Placing the paraventricular nucleus of the thalamus within the brain circuits that control
behavior. Neuroscience & Biobehavioral Reviews, 56, 315-329.

Kirouac, G. J. (2021). The paraventricular nucleus of the thalamus as an integrating and relay node in the
brain anxiety network. Frontiers in Behavioral Neuroscience, 15, 21.

Klein, A. S., Dolensek, N., Weiand, C., & Gogolla, N. (2021). Fear balance is maintained by bodily feedback
to the insular cortex in mice. Science, 374(6570), 1010-1015.

LeDoux, J. E. (2000). Emotion circuits in the brain. Annual Review of Neuroscience, 23(1), 155-184.

Letzkus, J. J., Wolff, S. B., Meyer, E. M., Tovote, P., Courtin, J., Herry, C., & Lüthi, A. (2011). A disinhibitory
microcircuit for associative fear learning in the auditory cortex. Nature, 480(7377), 331-335.

9
Likhtik, E., Stujenske, J. M., Topiwala, M. A., Harris, A. Z., & Gordon, J. A. (2014). Prefrontal entrainment of
amygdala activity signals safety in learned fear and innate anxiety. Nature Neuroscience, 17(1), 106-113.

Maren, S. (2001). Neurobiology of Pavlovian fear conditioning. Annual Review of Neuroscience, 24(1), 897-
931.

Murty, D. V., Song, S., Morrow, K., Kim, J., Hu, K., & Pessoa, L. (2022). Distributed and Multifaceted Effects
of Threat and Safety. Journal of Cognitive Neuroscience, 34(3), 495-516.

Nelson, A. B., & Kreitzer, A. C. (2014). Reassessing models of basal ganglia function and dysfunction. Annual
Review of Neuroscience, 37, 117.

Paré, D., & Quirk, G. J. (2017). When scientific paradigms lead to tunnel vision: lessons from the study of
fear. npj Science of Learning, 2(1), 1-8.

Paulus, M. P., & Stein, M. B. (2006). An insular view of anxiety. Biological Psychiatry, 60(4), 383-387.

Pessoa, L. (2014). Understanding brain networks and brain organization. Physics of Life Reviews, 11(3), 400-
435.

Pessoa, L. (2022). The Entangled Brain. Journal of Cognitive Neuroscience, 1-12.

Pessoa, L., Medina, L., Hof, P. R., & Desfilis, E. (2019). Neural architecture of the vertebrate brain:
implications for the interaction between emotion and cognition. Neuroscience & Biobehavioral Reviews, 107,
296-312.

Pessoa, L., Medina, L., & Desfilis, E. (2022). Refocusing neuroscience: moving away from mental categories
and towards complex behaviours. Philosophical Transactions of the Royal Society B, 377(1844), 20200534.

Shackman, A. J., & Fox, A. S. (2016). Contributions of the central extended amygdala to fear and
anxietycontributions of the central extended amygdala to fear and anxiety. Journal of Neuroscience, 36(31),
8050-8063.

Tovote, P., Fadok, J. P., & Lüthi, A. (2015). Neuronal circuits for fear and anxiety. Nature Reviews
Neuroscience, 16(6), 317-331.

Unzai, T., Kuramoto, E., Kaneko, T., and Fujiyama, F. (2015). Quantitative analyses of the projection of
individual neurons from the midline thalamic nuclei to the striosome and matrix compartments of the rat
striatum. Cereb. Cortex 27, 1164– 1181.

10
Yin, H. H., & Knowlton, B. J. (2006). The role of the basal ganglia in habit formation. Nature Reviews
Neuroscience, 7(6), 464-476.

11