In Review

Anxiety: An Evolutionary Approach

Melissa Bateson, DPhil1; Ben Brilot, PhD2; Daniel Nettle, PhD3

Anxiety disorders are among the most common mental illnesses, with huge attendant
suffering. Current treatments are not universally effective, suggesting that a deeper
understanding of the causes of anxiety is needed. To understand anxiety disorders better,
it is first necessary to understand the normal anxiety response. This entails considering its
evolutionary function as well as the mechanisms underlying it. We argue that the function of
the human anxiety response, and homologues in other species, is to prepare the individual to
detect and deal with threats. We use a signal detection framework to show that the threshold
for expressing the anxiety response ought to vary with the probability of threats occurring,
and the individual’s vulnerability to them if they do occur. These predictions are consistent
with major patterns in the epidemiology of anxiety. Implications for research and treatment are
discussed.

Can J Psychiatry. 2011;56(12):707–715.

Clinical Implications
• Understanding the evolved function of the anxiety response can aid in
understanding why people’s propensity to become anxious varies with their life
circumstances and developmental history.
• Theory predicts that the threshold for mounting an anxiety response should depend
on the probability of dangerous events occurring in the current environment,
and the vulnerability of the person to those events should they occur. Numerous
epidemiologic findings relating to anxiety disorders can be integrated within this
framework.
Limitations
• The evolutionary criteria for when a mechanism is functioning adaptively are
different from the criteria for identifying when a psychiatric disorder is present.
• Evolutionary thinking has not yet been used to develop new treatment strategies
for anxiety disorders, although it may aid in understanding why existing ones are
effective.

Key Words: anxiety, anxiety disorders, emotions, signal detection theory,

behavioural ecology, evolutionary medicine

It is not informative to study variations of need to be treated with the drugs to produce 1 more clinically
behaviour unless we know beforehand the norm significant symptomatic improvement than placebo.4 Only
from which the variants depart.1, p 109 around 46% of GAD patients show a clinically significant
response to psychological therapy.5 The pervasiveness of

A nxiety disorders are among the most common mental

illnesses, with huge attendant quality of life and financial
costs. For example, during 12 months in 2001–2002, 4.6%
anxiety disorders and their resistance to treatment suggest
a need for a deeper understanding of the sources of anxiety.
We argue that to make some progress understanding clinical
of Canadians had symptoms meeting DSM-IV diagnostic anxiety disorders, we need to start by obtaining a better
criteria for an anxiety disorder,2 and, during the course of understanding of the normal anxiety response: What is
their lifetimes, around 28.8% of Americans are estimated it for, and why does it have the features that it does? As
to be affected.3 Although treatments for anxiety exist, these evolution by natural selection is the source of all complex
currently have limited efficacy. For example, for GAD, physiological—including neural—organization in nature,
antidepressants, though more effective than placebo, have this necessarily means taking an evolutionary perspective.
a number needed to treat of 5.54, meaning nearly 6 patients Our concern is specifically with generalized anxiety.

The Canadian Journal of Psychiatry, Vol 56, No 12, December 2011 W 707
In Review

Anxiety is highly comorbid with depression, but we do systems are implicated). While generally accepted that
not have the space to explore evolutionary approaches to anxiety serves to prepare a person for threats,14 so far
depression (see instead Dr Hagen’s In Review paper6) or only a few pioneers have thought in any detail about the
why their comorbidity would be so high. We also do not implications of this being anxiety’s function.15,16 We believe
review the extensive evolutionary literature on specific that there may be considerable benefit for clinicians and
phobias.7 researchers in thinking more functionally, as this will shed
light on the interconnectedness of the different components
Why Do People Become Anxious? of anxiety and help guide and integrate research into the
Behavioural biologists divide explanatory questions into 2 brain mechanisms involved.
broad categories: those concerning ultimate evolutionary
function and those concerning proximate mechanisms.8,9 Are Some Cases of Anxiety Disorder Adaptive?
The ultimate function of a structure refers to the reasons that One question that tends to arise when considering psychiatric
the structure in its current form has been retained through conditions from an evolutionary point of view is where the
evolutionary time, against the many alternative forms of the boundary lies between evolved function and dysfunction.
same structure that are thrown up each generation through A mechanism is working functionally in the evolutionary
genetic mutation and recombination. For example, the sense if it has a level of responsiveness that will, averaged
ultimate function of the pigmentation in the skin of humans across all individuals and the environments in which they
is to protect tissue from the damaging effects of ultraviolet live, maximize survival and reproduction. This is a very
light. This claim is not a just-so story, because it makes different criterion from those used to demarcate clinical
principled predictions that can be tested in contemporary boundaries in psychiatry, which are mainly based on level
humans. It predicts that people whose skin colour is at the of suffering and quality of life. If a mechanism is producing
lightest end of the normal range of variation will suffer distress or impairing quality of life, this does not necessarily
increased rates of cancers and other ultraviolet-related health mean that it is malfunctioning in the evolutionary sense.
problems, such as neural tube defects in their offspring. For many adaptations, such as the pain system, it is part
These predictions are consistently confirmed.10–12 Thus, of their design that they cause subjectively unpleasant
given that these serious health problems affect survival and states, and individuals’ viability would be reduced if they
reproduction, we can state with some confidence that this did not do so under the appropriate circumstances. Thus,
account of the function of skin pigmentation is the correct while undoubtedly some cases of anxiety disorder are
one. pathological, in that the control mechanisms regulating
Mechanistic questions concern how a structure works. the anxiety response have become dysregulated, it is also
Human skin colour is produced by the pigment melanin, possible that some cases represent appropriate adaptive
which is synthesized from the amino acid tyrosine in responses to the situation in which the person currently finds
specialized cells called melanocytes, and transported to him or herself. As a corollary, insufficient anxiety-proneness
other epidermal cells in vesicles called melanosomes. may be a commonly occurring dysfunction. However, as it
These are all details of how pigmentation fulfills its is not associated with subjective distress, presumably the
function. Functional and mechanistic explanations can be people affected do not present for treatment.
pursued somewhat independently of one another in the Anxiety-proneness and anxious symptoms are distributed
first instance. We can work out what pigmentation is for along a continuum in the human population, and symptom
without necessarily knowing all the details of how melanin levels predict outcomes in a graded fashion.17,18 Moreover,
is produced. However, the 2 must ultimately be integrated.13 epidemiologic evidence suggests that the probability of long-
Functional thinking is potentially useful, even for the most term survival is lower in people with a low level of anxiety-
mechanistically oriented researchers, as the function of a proneness than those in the middle of the distribution.19,20
structure places constraints on how it can work and thus It is not even well established that, in people with clinical
limits the search space of possible hypotheses researchers levels of anxiety, the anxiety necessarily impairs biological
may want to pursue. fitness. In a recent study,20 patients with clinically defined
In psychiatric research, the focus has largely been on comorbid anxiety and depression had lower mortality than
mechanistic explanations for symptoms of anxiety (for those with depression alone, despite having poorer health
example, which drugs have anxiolytic or anxiogenic and more disability. All of this supports the view that clinical
functions, and which brain areas and neurophysiological anxiety shares a continuum with the normal, protective
anxiety response, and that locating the boundary between
adaptive function and pathology is not straightforward,
Abbreviations either philosophically or empirically. Moreover, the
presence of dysfunction may not be the appropriate criterion
ALR anxiety-like response
for allocating treatment (for discussion, see Wakefield21
DSM Diagnostic and Statistical Manual of Mental Disorders and Cosmides and Tooby22). In view of this continuity, it is
GAD generalized anxiety disorder important to consider in detail the evolutionary forces that
have shaped the anxiety response.

708 W La Revue canadienne de psychiatrie, vol 56, no 12, décembre 2011

 Anxiety: An Evolutionary Approach

Figure 1 The signal detection problem

Threshold Threshold

Background Background
noise Threat noise Threat

Misses False alarms Misses False alarms

The horizontal axes represent the strength of sensory information reaching the subject, while the 2 normal distributions represent the
probability of different amounts of evidence generated under the presence of a threat, and under background noise only. Crucially, the
2 distributions overlap. If the subject sets a high threshold for mobilizing a threat response (left panel), there will be a fair number of real
threats missed as well as a fair number of false alarms. If the subject sets a lower threshold (right panel), the subject will miss many fewer
real threats, but only at the cost of much more frequent false alarms. We argue that anxiety disorders represent unusually low personal
thresholds for threat response.

Function of ALRs in Other Species the ALR prepares the body for action by hyperventilating
Evolutionary thinking in behavioural biology provides the to oxygenate the blood, diverting blood to the muscles
researcher with a toolbox of different approaches. We will and sweating to cool the skin. Thus both the contexts in
employ 2: the comparative and the optimality approaches. which anxious symptoms appear and the nature of these
symptoms support the view that the function of anxiety is
Comparative Evidence to prepare, physiologically, cognitively, and behaviourally,
The comparative approach is based on comparing different for detecting and dealing with threats to survival.
species, or different populations of the same species, to
discover the ecological correlates of a behaviour pattern, An Optimality Approach
and thereby to test hypotheses about function. Psychiatrists Optimality approaches in behavioural biology involve
have long employed comparative evidence from other building theoretical models of what behaviour we ought
species when thinking about mechanisms of anxiety (most to expect if the structure has indeed been selected to fulfil
psychopharmacology is done initially in animal models), the function we hypothesize (for an example in the context
but there is scope for greater use of comparative evidence of low mood, see Nettle27). Model predictions can then be
in also investigating function. compared to actual data on the behaviour in question. We
have already suggested that the function of anxiety is to
Animals of many species display a suite of responses that we
prepare the individual for threats. How should such a threat
call the ALR in their normal behaviour. The ALR consists of
detection system be designed? Nesse28,29 introduced the
increased heart rate, stress hormone secretion, restlessness,
use of signal detection theory to address this problem, an
vigilance, and fear of potentially dangerous environments,
approach we extend here.
and decreased feeding and exploratory behaviour. It also
involves an increased tendency to interpret ambiguous Signal detection theory is a body of mathematics concerned
stimuli as threatening.23–25 In Belding’s ground squirrels with the problem of how to decide whether a given event
(Spermophilus beldingi), the components of the ALR are has occurred (here, the event is whether a threat is present;
more strongly expressed in populations facing higher levels for example, was that rustle in the bushes a tiger or only
of predation.26 Differences in stress hormones in high- and the wind?). Ideally, sensory evidence would indicate this
low-predation environments are detectable within a few unambiguously, but, in practice, there is fluctuating ambient
weeks of the pups emerging from the nest. This suggests noise; although, on average, tigers make more rustling than
that the function of the ALR is to detect and deal with the wind, in some instances a tiger generates less rustling
threats, specifically from predators in this case. The changes than in other instances where there is no tiger. Therefore,
in physiology, cognition, and behaviour characterizing the there are overlapping distributions of cue intensity reaching
ALR can be interpreted as ways of fulfilling this function. the receiver from environments with tigers and from safe
The cognitive changes increase the likelihood that imminent environments (Figure 1). The receiver must establish some
threats will be detected early, and, at the physiological level, criterion (called the signal detection threshold) above

The Canadian Journal of Psychiatry, Vol 56, No 12, December 2011 W 709
In Review

which they will accept the evidence as indicating that the Figure 2 The optimal threshold for mounting a threat
event is present (here, a level of rustling they will accept as response, as a function of the probability of that threat being
present in the environment, for individuals with 3 levels of
sufficient to prepare for a tiger). Below this threshold, the
vulnerability
rustling is dismissed as noise. Because the 2 distributions
in Figure 1 overlap, any threshold that the receiver sets
will generate some errors; it is impossible to set a threshold
where the outcomes are all correct, whether detections (hits)
or rejections. Specifically, there will always be some false
alarms, where the receiver prepares for a tiger when it turns
out to be nothing but the wind, and some misses, where the
receiver does not prepare but there is a tiger.
The optimization problem concerns where to set the
threshold. In general, there is a direct trade-off between the
frequencies of false alarms and misses; raising the threshold
will reduce the former but only at the expense of increasing
the latter. The optimal threshold (λ) for the receiver to adopt
can be derived mathematically,30 and can be simplified for
our purposes as follows in Equation 1:

pnt w fa
m= :
pt wmiss
Values are generated from Equation 1, with the vulnerability
term set at 0.1, 0.33, or 1.0. Increasing the probability of
where pnt and pt are the probabilities of there being no threatening events occurring always lowers the optimal threshold
threat present, and a threat present, respectively, and wfa and for responding, as does increasing the individual’s personal
wmiss are the respective costs of a false alarm and a miss vulnerability to threats should they occur.
(see also Haselton and Nettle31). Note that costs here are in
terms of biological fitness. Thus, if misses usually lead to
The second prediction is that the relative costs of false
death, wmiss is very large. The pnt/pt term is determined by
alarms and misses (that is, the vulnerability) will affect the
the objective chance that a threat is imminent (that is, the
threshold. If misses are generally costly, as seems reasonable
threat probability). We call the wfa/wmiss term the individual’s
vulnerability, as it specifies how bad it would be for that for the detection of mortal dangers (“few failures are as
individual’s fitness if a real threat went undetected. unforgiving as failure to avoid a predator”32, p 619) and the
cost of a false alarm is only some unnecessary vigilance and
A couple of predictions follow from this equation increased heart rate, then an optimally functioning system
(summarized in Figure 2). First, as the actual prevalence of will produce many times more false alarms than misses.
threats in the environment goes up (that is, the probability This is Nesse’s smoke detector principle28,29; when they
increases), the optimal threshold for mounting the ALR are set correctly, smoke detectors go off when there is no
gets lower. This means that when an optimally behaving fire moderately often, but never miss a real fire. Setting the
animal is in an environment where there is objectively a lot threshold higher may be nicer for residents of the building,
of danger (or where it has received prior cues that dangers who would not have to troop outside for no reason so often,
are frequent), it should require much less evidence of the but in the longer term it would be suboptimal, as one missed
presence of a threat to trigger the ALR than it would when fire is a catastrophe.
in a safe environment. This means we should predict that
animals will display many more anxiety symptoms in More specifically, though, any factor which affects the
such environments. Some of the additional ALRs will be individual’s ability to cope with the danger, and which
hits, owing to the greater prevalence of threats, but many thereby increases the likely cost of a miss for that individual,
will be false positives, as lowering the detection threshold ought to cause that individual to lower their threshold
necessarily entails increasing the false alarm rate. Figure 2 accordingly. From Figure 2, we can see that an individual
shows that as probability increases, the optimal threshold with higher vulnerability has a lower optimal threshold
decreases initially at a faster than linear rate, which means for any given level of probability than a less vulnerable
that as the hit rate goes up, the false alarm rate goes up individual. For example, this means that animals ought to
even faster. Thus, as the individual’s best estimate of the have a lower threshold for mounting an anxiety response
amount of danger around goes up, the amount of both when they are lame than when they are healthy, and when
correct and baseless threat responses ought to increase. We they are in an environment with no cover, compared with
have already seen that this prediction holds in the case of one with cover. There is certainly evidence for such effects,
Belding’s ground squirrels. both from the laboratory and the field. In the laboratory,

710 W La Revue canadienne de psychiatrie, vol 56, no 12, décembre 2011

 Anxiety: An Evolutionary Approach

Table 1 Symptoms of anxiety and their function in defence Application of the Evolutionary Approach to
against potential threats Anxiety in Humans
Symptom Functional significance Let us assume that human anxiety is the homologue of the
Easily startled, hypersensitive Response to threat easily ALR found in many other species and that it serves the
to noise evoked same function, to detect and prepare for threats. How does
Insomnia Constant alertness what we know from the psychiatric literature fit in with this
Restlessness, increased heart Body is prepared for action assumption?
rate
Preferential attention to cues Notice threats sooner Implications for Symptoms of Anxiety
related to threats Many symptoms of human anxiety, although unpleasant,
Interpretation of ambiguous Reduce probability of missing make functional sense in the terms we have outlined (Table
information as threatening possible threats 1). At the cognitive level, vigilance is increased, threatening
Ambiguity aversion Avoidance of situations whose information is given processing priority,39 and ambiguous
threat level is unclear information is interpreted as threatening40 because threats
are judged as more likely to occur41,42 (though see Nesse and
Klaas43). The functional interpretation of these changes is
obvious. We make a more specific claim: anxiety disorders,
placing animals in isolation or exposed settings produces
specially, GAD, can be equated to having a low threshold
increased ALR, including cognitive changes toward
in the signal detection model. This makes sense of many
interpreting ambiguous stimuli as negative.24,33 In the field,
GAD features, which are basically extreme forms of the
individuals who lack cover, are far from conspecifics or at
normal anxiety response features outlined in Table 1. The
the margins of social groups, show increased vigilance.34,35
DSM-IV specifies that the anxiety must be characterized as
Pregnant and nursing females are also more vigilant.36,37
excessive for a period of at least 6 months.44 The excessive
Presumably these effects are due to individual vulnerability
in this condition is hard to operationalize exactly, but
in the event of an attack being elevated. Additionally, in reef
it could be taken to mean the presence of threat-related
fish Chromis dimidiata and Pseudanthias squamipinnis,
cognition with an unusually high proportion of false alarms.
individuals who have had access to a cleaner wrasse have
Such a high proportion follows directly from having a low
an attenuated stress response, compared with individuals
threshold. The other GAD symptoms—for example, sleep
who have not.38 The interpretation of this finding is that
difficulties, tension, and poor concentration—all follow
cleaner wrasse, by removing ectoparasites, keep the fish
straightforwardly from the person’s threshold for threat
in better overall health. Better health means their ability
detection being set so low that almost any information from
to cope with subsequent threats is greater, and thus, in our
the external environment becomes significant to provoke a
terms, their vulnerability is lower. Thus it makes sense that
response. We side-step the question of whether the threshold
their threshold for mounting an ALR would be higher.
is low owing to system malfunction—pathology—or
In summary, we expect an ALR optimized by natural that being the best threshold for that person’s current
selection to exhibit the following features. life situation. No doubt it varies from person to person.
1. Its ease of evocation would not be fixed but rather Moreover, we ought to expect there to be both traitlike (that
modulated, depending on the individual’s environment is, temperamental) variation in where people’s thresholds
and current state. are, and within-individual change in thresholds according
2. It would be more readily evoked in environments to their current life situations. Nonetheless, GAD’s suite of
where threats are more common (that is, where the symptoms is coherent and can be linked to having a low
probability is high). threshold in the signal detection model.
3. It would be more readily evoked when individuals
are in a condition that means they will find it difficult Implications for Epidemiology
to cope with undetected threats (that is, where the If anxiety has the functions suggested for the ALR in our
vulnerability is high). section Function of ALRs in Other Species, and if the signal
4. It will consist of a suite of physiological, cognitive, detection model is a good framework for understanding
and behavioural changes that divert the individual’s its design, we can make clear predictions about the
resources from other concerns, and facilitate the epidemiology of anxiety. These really all come down to
detection of and response to threats. 2 claims: as the objective probability of threats increases,
5. Even when functioning optimally, the system will so should the severity of anxious symptoms, and as the
produce many more false alarms than hits, and the vulnerability of people to threats, should they occur, goes
greater the probability and vulnerability, the greater up, so should the severity of anxious symptoms. Many of
the ratio of false alarms to hits will be. the specific epidemiology patterns observed in the literature
There is some support for at least the first 3 of these can be related to these 2 claims (Table 2). We find the
propositions from nonhuman examples. To what extent, evolutionary framework useful in that a large number of
though, do they help us understand anxiety in humans? empirically derived associations presented in the literature

The Canadian Journal of Psychiatry, Vol 56, No 12, December 2011 W 711
In Review

Table 2 Epidemiologic associations with anxiety that can be interpreted from the standpoint of the signal detection model
Pattern Reference Interpretation
Low birth weight predicts higher trait anxiety later in Lahti et al45 Higher vulnerability: low birth weight
life individuals in poorer somatic state and have
higher lifelong vulnerability to mortality and
morbidity
Low socioeconomic position associated with greater Najman et al46 and Higher probability and vulnerability: low
anxiety McMillan et al47 socioeconomic position characterized by
more bad events, and fewer resources to
cope with them when they occur
Injuries or disabilities causing loss of physical Cano et al,49 Suh et al,50 Higher vulnerability: loss of mobility reduces
mobility associated with increased anxiety (and this Lenze et al,51 and Bellin et al52 people’s ability to deal with threatening
seems to be independent of the association between situations
depression and physical conditions48)
Members of discriminated-against minorities have Das-Munshi et al53 Higher probability: probability of threats
less anxiety when living in neighbourhoods with higher for minority individuals suffering
more people of the same ethnicity discrimination when there are few others
nearby
High absolute levels of anxiety in Zimbabwe, Central Langhaug et al,54 Vinck and Higher probability: high rates of violence,
African Republic, and Gaza Strip Pham,55 Elbedour et al56 unrest and disease mean probability of
mortal threat is high in these environments
More anxiety among women than men Breslau et al57 Higher probability and (or) vulnerability:
greater level of threat and (or) vulnerability
to threats among women than men
Anxiety associated with living alone or being a lone Rimehaug and Wallander58 Higher vulnerability: threats more difficult to
parent cope with in the absence of social support
More anxiety among women suffering domestic Tolman and Rosen59 Higher probability: prevalence of physical
violence danger is higher for such women

as unconnected all emerge from the general framework in never experience it responding with threatening behaviour
a coherent way. Some of these are obvious and would have (“repeated disconfirmation” is a term used to describe the
been predicted anyway, but others, such as the strong effect therapeutic methodology in this source”), then their best
of loss of physical mobility on anxiety, are not intuitive estimate of the probability term in Equation 1 for the case of
until one adopts the evolutionary view. dogs will be much lower. Their threshold for the evocation
of anxious responses when dogs are in the environment
Implications for Therapy should thus be raised, and symptomatic relief follows.
We believe that the framework outlined here could be Drug treatments for anxiety disorders work by down-
useful for developing novel approaches to treatment, or at regulating threat detection mechanisms pharmaco­logically.
least refining existing ones. The effects of actual levels of An interesting possibility raised by our analysis is that
probability and vulnerability must be mediated by a person’s drugs that treat nonanxiety-related symptoms that feed into
appraisal (conscious or otherwise) of their probability perceived vulnerability, such as, for example, providing
and vulnerability.60 This means that relief of anxiety can, good analgesia to someone with chronic pain, may also
in principle, come in 2 ways; reducing people’s objective reduce symptoms of anxiety.
probability and vulnerability, and changing their appraisals
The perspective outlined here draws attention to the fact
thereof, which may or may not be realistic (for a related
that at least some people may be anxious because they
discussion, see Nesse and Ellsworth61).
are correct to be anxious. That is, their appraisals of high
Some existing therapies can work by changing appraisals of probability and vulnerability may be relatively realistic, and
probability–vulnerability. Cognitive therapies62 for anxiety their symptoms, although unpleasant, reflect an adaptive
may address either the vulnerability or the probability response to this situation. Thus, on a holistic view, it is
term of Equation 1. Recalibrating either probability or a priority to address ecological issues, such as poverty,
vulnerability ought to result in raising the threshold for economic insecurity, lack of social support, poor housing,
evocation of anxious symptoms. Exposure therapy is a and urban safety, as well as challenging people’s cognition.
long-established and often effective framework for treating These ecological factors are strongly associated with mental
specific phobias, which involves the controlled exposure distress at both the individual and the societal level.64
of the patient to the source of the anxiety.63 Exposure Tackling these factors will shift the population distributions
therapy may be having its effect by recalibrating the of experienced probabilities of and vulnerabilities to threats,
person’s estimate of the probability of a specific threat and thus ought to have population-level benefits regarding
occurring; if the patient can handle a dog many times, and lower rates of anxiety.

712 W La Revue canadienne de psychiatrie, vol 56, no 12, décembre 2011

 Anxiety: An Evolutionary Approach

If we accept that some cases of clinical anxiety may represent References

adaptive responses to the situations in which people find 1. Medawar P. The art of the soluble. London (GB): Methuen; 1967.
themselves, this raises an important question about the 2. Tempier R, Meadows GN, Vasiliadis HM, et al. Mental disorders
and mental health care in Canada and Australia: comparative
consequences of treatment. If the treatment involves
epidemiological findings. Soc Psychiatry Psychiatr Epidemiol.
reduction of actual levels of probability or vulnerability, 2009;44(1):63–72.
this is fine, because reduced anxiety is appropriate in these 3. Kessler RC, Berglund P, Demler O, et al. Lifetime prevalence
circumstances. However, if cognitive or pharmacological and age-of-onset distributions’ of DSM-IV disorders in the
treatment involves a change in a person’s appraisal of the national comorbidity survey replication. Arch Gen Psychiatry.
levels of probability or vulnerability without a change in the 2005;62(6):593–602.
4. Kapczinski F, Lima MS, Souza JS, et al. Antidepressants for
real levels, then there is a danger that this could increase the
generalized anxiety disorder. Cochrane Database Syst Rev.
chances of a miss, with dire consequences for the person. 2003;(2):CD003592.
Removing the battery from a smoke detector may seem like 5. Hunot V, Churchill R, Silva de Lima M, et al. Psychological
a good solution to frequent false alarms, until a real fire therapies for generalised anxiety disorder. Cochrane Database Syst
hits. The very fact that evidence is starting to accumulate Rev. 2007;(1):CD001848.
6. Hagen EH. Evolutionary theories of depression: a critical review.
that anxious people are less likely to suffer various forms
Can J Psychiatry. 2011;56(12):716–726.
of mortality should at least make us question whether there 7. Ohman A, Mineka S. Fears, phobias, and preparedness: toward
may sometimes be huge costs to inappropriately treating an evolved module of fear and fear learning. Psychol Rev.
anxiety disorders. 2001;108(3):483–522.
8. Tinbergen N. [On aims and methods in ethology]. Z Tierpsychol.
1963;20:410–433. [Article in German].
9. Scott-Phillips TC, Dickins TE, West SA. Evolutionary theory and
Conclusions the ultimate–proximate distinction in the human behavioral sciences.
We have outlined a simple functional framework for Perspect Psychol Sci. 2011;6:38–47.
understanding human anxiety and homologous responses 10. Lear JT, Tan BB, Smith AG, et al. Risk factors for basal cell
in other species. An individual’s threshold for mounting a carcinoma in the UK: case–control study in 806 patients. J R Soc
threat response ought to be sensitive to the likelihood of bad Med. 1997;90(7):371–374.
11. Stevenson RE, Allen WP, Pai GS, et al. Decline in prevalence
things happening in the environment (probability), and the
of neural tube defects in a high-risk region of the United States.
individual’s ability to cope if they do happen (vulnerability). Pediatrics. 2000;106(4):677–683.
If either increases, the individual anxiety response should 12. Okoro AN. Albinism in Nigeria. A clinical and social study. Br J
be more easily triggered, and should therefore produce a Dermatol. 1975;92:485–492.
greater number of false alarms. Epidemiologic patterns of 13. McNamara JM, Houston AI. Integrating function and mechanism.
anxiety symptoms can be interpreted within this framework, Trends Ecol Evol. 2009; 24(12):670–675.
14. Beck AT, Emery G, Greenberg RL. Anxiety disorders and phobias:
and existing treatments for anxiety disorders can be a cognitive perspective. New York (NY): Basic Books; 1985.
understood as reducing the patient’s appraisal of either their 15. Marks IM, Nesse RM. Fear and fitness—an evolutionary analysis of
probability or vulnerability. anxiety disorders. Ethol Sociobiol. 1994;15(5–6):247–261.
16. Hinds AL, Woody EZ, Drandic A, et al. The psychology of potential
The evolutionary perspective we have outlined does not
threat: properties of the security motivation system. Biol Psychol.
immediately change what psychiatrists do. However, it does 2010;85(2):331–337.
bring a broader biological lens to bear on a complex issue 17. Kessler RC, Merikangas KR, Berglund P, et al. Mild disorders
that is associated with a huge burden of human suffering. should not be eliminated from the DSM-V. Arch Gen Psychiatry.
It raises the potential for integration between psychiatric 2003;60(11):1117–1122.
18. Angst J, Gamma A, Baldwin DS, et al. The generalized anxiety
research and the rich traditions of research in behavioural
spectrum: prevalence, onset, course and outcome. Eur Arch
ecology and ethology. It may lead to fresh, more holistic Psychiatry Clin Neurosci. 2009;259(1):37–45.
public health approaches to anxiety prevention, and 19. Lee WE, Wadsworth MEJ, Hotopf A. The protective role
stimulate further research by promoting the exchange of of trait anxiety: a longitudinal cohort study. Psychol Med.
ideas between scientists working on research topics as 2006;36(3):345–351.
diverse as epidemiology, neurobiology, clinical psychology, 20. Mykletun A, Bjerkeset O, Overland S, et al. Levels of anxiety
and depression as predictors of mortality: the HUNT study. Br J
and animal behaviour. In particular, thinking in evolutionary
Psychiatry. 2009;195(2):118–125.
terms draws attention to the way high levels of anxiety may 21. Wakefield JC. The concept of mental disorder—on the boundary
be associated not only with neural malfunction but also between biological facts and social values. Am Psychologist.
with genuinely dangerous or precarious life situations. That 1992;47(3):373–388.
is, to understand behaviour, one must consider what is not 22. Cosmides L, Tooby J. Toward an evolutionary taxonomy of treatable
conditions. J Abnorm Psychol. 1999;108(3):453–464.
only inside the organism but also going on in the organism’s
23. Brilot BO, Asher L, Bateson M. Stereotyping starlings are more
immediate environment. ‘pessimistic’. Anim Cogn. 2010;13(5):721–731.
24. Salmeto AL, Hymel KA, Carpenter EC, et al. Cognitive bias in the
chick anxiety-depression model. Brain Res. 2011;1373:124–130.
Acknowledgement 25. Burman OHP, Parker RMA, Paul ES, et al. Anxiety-induced
The Canadian Psychiatric Association proudly supports the cognitive bias in non-human animals. Physiol Behav.
In Review series by providing an honorarium to the authors. 2009;98(3):345–350.

The Canadian Journal of Psychiatry, Vol 56, No 12, December 2011 W 713
In Review

26. Mateo JM. Ecological and hormonal correlates of antipredator 48. Scott KM, Bruffaerts R, Tsang A, et al. Depression-anxiety
behavior in adult Belding’s ground squirrels (Spermophilus relationships with chronic physical conditions: results
beldingi). Behav Ecol Sociobiol. 2007;62(1):37–49. from the World Mental Health Surveys. J Affect Disord.
27. Nettle D. An evolutionary model of low mood states. J Theor Biol. 2007;103(1–3):113–120.
2009;257(1):100–103. 49. Cano LA, Zafra AO, Toro EO. [Effect of injuries on psychological
28. Nesse RM. Natural selection and the regulation of defenses: a signal variables among under-19 soccer players]. Rev Latinoam Psicol.
detection analysis of the smoke-detector problem. Evol Hum Behav. 2010;42(2):265–277. [Article in Spanish].
2005;26:88–105. 50. Suh Y, Motl RW, Mohr DC. Physical activity, disability, and
29. Nesse RM. The smoke detector principle. Natural selection mood in the early stage of multiple sclerosis. Disabil Health J.
and the regulation of defensive responses. Ann N Y Acad Sci. 2010;3(2):93–98.
2001;935:75–85. 51. Lenze EJ, Rogers JC, Martire LM, et al. The association of
30. Green DM, Swets JA. Signal detection theory and psychophysics. late-life depression and anxiety with physical disability: a review
New York (NY): John Wiley & Sons, Ltd; 1966. of the literature and prospectus for future research. Am J Geriatr
31. Haselton MG, Nettle D. The paranoid optimist: an integrative Psychiatry. 2001;9(2):113–135.
52. Bellin MH, Zabel TA, Dicianno BE, et al. Correlates of depressive
evolutionary model of cognitive biases. Pers Soc Psychol Rev.
and anxiety symptoms in young adults with spina bifida. J Pediatr
2006;10:47–66.
Psychol. 2010;35(7):778–789.
32. Lima SL, Dill LM. Behavioural decisions made under the
53. Das-Munshi J, Becares L, Dewey ME, et al. Understanding the
risk of predation—a review and prospectus. Can J Zool.
effect of ethnic density on mental health: multi-level investigation of
1990;68(4):619–640.
survey data from England. BMJ. 2010;341c5367.
33. Bateson M, Matheson SM. Performance on a categorisation
doi:10.1136/bmj.c5367.
task suggests that removal of environmental enrichment induces
54. Langhaug LF, Pascoe SJ, Mavhu W, et al. High prevalence of
‘pessimism’ in captive European starlings (Sturnus vulgaris). Anim affective disorders among adolescents living in Rural Zimbabwe.
Welf. 2007;16:33–36. J Community Health. 2010;35(4):355–364.
34. Devereux CL, Whittingham MJ, Fernández-Juricic E, et al. Predator 55. Vinck P, Pham PN. Association of exposure to violence and potential
detection and avoidance by starlings under differing scenarios of traumatic events with self-reported physical and mental health status
predation risk. Behav Ecol. 2006;17(2):303–309. in the Central African Republic. JAMA. 2010;304(5):544–552.
35. Quenette PY. Functions of vigilance behaviour in 56. Elbedour S, Onwuegbuzie AJ, Ghannam J, et al. Post-traumatic
mammals—a review. Acta Oecol Int J Ecol. 1990;11(6):801–818. stress disorder, depression, and anxiety among Gaza Strip
36. Pangle WM, Holekamp KE. Functions of vigilance behaviour in a adolescents in the wake of the second Uprising (Intifada). Child
social carnivore, the spotted hyaena, Crocuta crocuta. Anim Behav. Abuse Negl. 2007;31(7):719–729.
2010;80(2):257–267. 57. Breslau N, Schultz L, Peterson E. Sex differences in depression:
37. Monclús R, Rödel HG. Influence of different individual a role for preexisting anxiety. Psychiatry Res. 1995;58(1):1–12.
traits on vigilance behaviour in European rabbits. Ethology. 58. Rimehaug T, Wallander J. Anxiety and depressive symptoms related
2009;115(8):758–766. to parenthood in a large Norwegian community sample: the HUNT2
38. Bshary R, Oliveira RF, Oliveira TS, et al. Do cleaner organisms study. Soc Psychiatry Psychiatr Epidemiol. 2010;45(7):713–721.
reduce the stress response of client reef fish? Front Zool. 2007;4:21. 59. Tolman RM, Rosen D. Domestic violence in the lives of women
39. Mathews A, MacLeod C. Selective processing of threat cues in receiving welfare—mental health, substance dependence, and
anxiety states. Behav Res Ther. 1985;23(5):563–569. economic well-being. Violence Against Women. 2001;7(2):141–158.
40. Eysenck MW, Mogg K, May J, et al. Bias in interpretation of 60. Scherer KR, Schorr A, Johnstone T. Appraisal processes in emotion:
ambiguous sentences related to threat in anxiety. J Abnorm Psychol. theory, methods, research. New York (NY): Oxford University
1991;100(2):144–150. Press; 2001.
41. Butler G, Mathews A. Anticipatory anxiety and risk perception. 61. Nesse RM, Ellsworth PC. Evolution, emotions, and emotional
Cogn Ther Res. 1987;11(5):551–565. disorders. Am Psychol. 2009;64(2):129–139.
42. Butler G, Mathews A. Cognitive processes in anxiety. Adv Behav 62. Beck AT. Cognitive therapy and the emotional disorders. New York
Res Ther. 1983;5(1):51–62. (NY): International Universities Press; 1967.
43. Nesse RM, Klaas R. Risk perception by patients with anxiety 63. Rachman S. The modification of agoraphobic avoidance behaviour:
disorders. J Nerv Ment Dis. 1994;182:465–470. some fresh possibilities. Behav Res Ther. 1983;21(5):567–574.
64. Wilkinson RG, Pickett KE. Income inequality and population
44. American Psychiatric Association. Diagnostic and statistical manual
health: a review and explanation of the evidence. Soc Sci Med.
of mental disorders. 4th ed. Washington (DC): APA; 1994.
2006;62(7):1768–1784.
45. Lahti J, Räikkönen K, Pesonen AK, et al. Prenatal growth, postnatal
growth and trait anxiety in late adulthood—the Helsinki Birth
Cohort Study. Acta Psychiatr Scand. 2010;121(3):227–235. Manuscript received June 2011, revised, and accepted August 2011.
46. Najman JM, Hayatbakhsh MR, Clavarino A, et al. Family poverty 1
Reader in Ethology, Newcastle University, Newcastle, England.
over the early life course and recurrent adolescent and young adult 2
Research Associate, Newcastle University, Newcastle, England.
anxiety and depression: a longitudinal study. Am J Public Health. 3
Professor of Behavioural Science, Newcastle University, Newcastle,
2010;100(9):1719–1723. England.
47. McMillan KA, Enns MW, Asmundson GJ, et al. The association Address for correspondence: Dr Melissa Bateson, Centre for Behaviour
between income and distress, mental disorders, and suicidal and Evolution, Institute of Neuroscience, Newcastle University, Henry
ideation and attempts: findings from the collaborative psychiatric Wellcome Building, Framlington Place, Newcastle NE2 4HH, UK;
epidemiology surveys. J Clin Psychiatry. 2010;71(9):1168–1175. [Link]@[Link]

714 W La Revue canadienne de psychiatrie, vol 56, no 12, décembre 2011

 Anxiety: An Evolutionary Approach

Résumé : L’anxiété : une approche évolutionnaire

Les troubles anxieux comptent parmi les maladies mentales les plus répandues, et une
intense souffrance les accompagne. Les traitements actuels ne sont pas universellement
efficaces, ce qui suggère qu’il faut une compréhension plus profonde des causes
de l’anxiété. Pour mieux comprendre les troubles anxieux, il est d’abord nécessaire
de comprendre la réponse normale à l’anxiété, ce qui veut dire examiner sa fonction
évolutionnaire de même que les mécanismes qui la sous-tendent. Nous alléguons que
la fonction de la réponse humaine à l’anxiété, et des homologues d’autres espèces, est
de préparer la personne à détecter et à traiter les menaces. Nous utilisons un cadre de
détection des signaux pour montrer que le seuil d’expression de la réponse à l’anxiété
devrait varier selon la probabilité que des menaces se produisent, et selon la vulnérabilité
de la personne à ces menaces si elles sont présentes. Ces prédictions concordent avec les
principaux modèles de l’épidémiologie de l’anxiété. Les implications pour la recherche et le
traitement sont discutées.

The Canadian Journal of Psychiatry, Vol 56, No 12, December 2011 W 715