what is anxiety anyway?

I’ve been trying to think about anxiety recently.  I want to argue that anxiety is part of a broader meta-level control system involved in risk assessment.  The obvious scientific response to this would be something along the lines of “well, duh.”  But I’ve had a great deal of difficulty articulating exactly what I mean to my friends.  So I’m writing this overly long post to try to work through it.  I’ll try using two different analogies, but first I’ll try to give some motivating background.

Why it isn’t obvious

When I started out working on my current project, I intended to include a high-level simulation of Gray and McNaughton’s revised (2000) version of Reinforcement Sensitivity Theory.  RST is a neuropsychological theory of anxiety that has been influential in the personality theory community.  RST argues (in the revised version) that there are three main behavioral subsystems:

  • The Behavioral Activation System (BAS), which is responsible for approach behaviors,
  • The Fight/Flight/Freezing system (FFFS), which is responsible for avoidance behaviors, and
  • The Behavioral Inhibition System (BIS), which resolves conflicts between the two (or within one system).


  • Anxiety is triggered by goal conflicts detected by the BIS.
  • Anxiety is responded to by the BIS (1) directing attention toward the source of the conflict to gather more information, then (2) resolving the goal conflict by assessing the stimulus and inhibiting one option while stimulating another.  The system has a bias toward the FFFS, so it’s more likely to resolve the conflict through some kind of escape behavior than through approach.

Along the way, I got distracted by the Attachment Theory literature.  In attachment experiments:

  • Anxiety is triggered by the non-availability (physical or emotional) of the care-giver, unfamiliar environments, and the presence of strangers.
  • Anxiety is responded to by activation of the Attachment Behavior System, which causes the child to try to look to the caregiver, achieve eye contact with them, or get physically closer to them, depending on the level of anxiety.
  • (Note however, that Attachment Theory doesn’t claim these are the sole causes or responses to anxiety).

A securely attached child need not maintain proximity to the parent at all times, but uses the parent as a “secure home base” from which to make exploratory excursions.  Essentially, the child hangs out with the care-giver, charges up his/her security battery, then goes out and explores until the security battery gets depleted, then recharges, either by actually returning to the caregiver, by achieving eye contact with him/her, or simply by making sure the child still have line-of-sight to the care-giver.

The thing to notice here is that RST models anxiety as being by definition goal conflict.  Attachment theory doesn’t define anxiety but includes among the things that can trigger it, the inaccessibility of the care-giver.  But inaccessibility isn’t goal conflict.  It’s closer to goal failure, although the fact that child’s desire for parental availability waxes and wanes means it’s more complicated than a nice simple maintenance goal like Always(Near(parent)).

In other words, the neural basis for anxiety provided by RST probably isn’t the neural basis for what attachment theory calls anxiety.  So the question then is whether the phenomenon RST is calling anxiety and the one attachment theorists call anxiety are the same phenomenon, and if so, what the phenomenon “really” is.

An immune system analogy

Your immune system can be grossly divided into the specific and non-specific immune systems, (or adaptive and innate).  Each has a set of detection mechanisms and a set of response mechanisms.  The specific immune system detects threats using antibodies that recognize specific proteins on the threats and responds by (among other things) attacking whatever is attached to the naughty protein.  The non-specific system notices general signs that something is awry, such as chemical markers of (non-programmed) cell death, and responds with general strategies to discourage infection, such as fever and inflammation, as well as general sensitization of the rest of the immune system.

The fight/flight system is also essentially a system for dealing with specific threats; fight and flight are only possible when a specific aversive object has been identified to attack or avoid.  But there are a number of stimuli that aren’t threats themselves, but rather general cues that you might be in a more dangerous situation than you’re used to.  Examples of these would be:

  • Darkness
  • Unfamiliar environments, situations or people
  • Losing contact with a care-giver, or more generally, being alone
  • Displays of anxiety in those around you
  • The world behaving in ways it isn’t supposed to

Consequently, your only possible responses are to:

  • Activate safety-seeking behaviors, such as attachment, social engagement (the “tend and befriend” system), or withdrawl (hiding)
  • Increase your vigilance level
  • Adopt a risk-averse action-selection strategy, e.g. by biasing action selection toward the FFFS
  • Prepare for any concrete threats that might appear, e.g. by releasing adrenaline

Notice that both the stimuli and responses listed above are generally associated with anxiety.

One can argue that this works well for the kinds of anxiogenic stimuli considered in attachment theory, but it doesn’t work well for goal conflict.  But while goal conflict isn’t a cue to an increased risk of encountering threatening objects, it is a cue that you’re moving outside of situations you know how to handle easily; the BIS doesn’t fire for all goal conflicts, just goal conflicts you haven’t already learned to resolve.  So again, it’s a cue to possible future problems and so argues for shifting to a more conservative strategy.

a Vygotskian analogy

Vygotsky was a developmental psychologist who, among other things, developed the notion of the zone of proximal development.  The idea is that when you’re learning a domain, the tasks within it fall into those you’ve already mastered, those that are too damn hard, and those you don’t know yet but are ready to learn.  The latter group forms the zone of proximal development.  An instructor’s job is to keep the student in the ZPD by taking responsibility for the parts of the task that are currently too hard for the student to learn, a process known as scaffolding.   As the student learns more, the ZPD shifts, and the instructor can gradually remove the scaffolding.  (I’m a fan of this model in part because I think the reason programming is so hard to teach is that programming is very difficult to scaffold effectively).

The important concepts are that there’s:

  • A continuum of activity, from too hard to too easy, and
  • A set of scaffolding techniques that can be modulated to steer the learner toward a sweet spot in the continuum

However, there’s no reason the scaffolding need be done exclusively by a teacher.  To the extent that it’s possible to make the world seem harder or easier (at a cost) by shifting between risk seeking and risk aversion, or by controlling the level of dependence on a care-giver, we can think of the child as self-scaffolding.  In fact, we can even think of adults as self-scaffolding to the extent that they modulate these behaviors, even if they aren’t necessarily learning.  So we can think of anxiety as the feedback signal that’s telling the agent it needs to increase its level of scaffolding.

So in other words, the argument is that there’s a general meta-level process of security assessment that modulates both high-level parameters like risk aversion, and remediation behaviors like attachment.  It’s essentially a homeostatic process.  You want enough risk aversion and remediation to attain an acceptable level of security, but not so much that wasting resources preparing for threats that probably aren’t there.  Anxiety is the signal to increase the response of the security system, while other signals like boredom would tend to decrease its activity.

There.  Does that make sense?

Technorati Tags: , , ,


4 Responses to “what is anxiety anyway?”

  1. August 27, 2007 at 7:47 pm

    Really interesting ideas! However, I’m not sure they capture my intuitive sense of anxiety.

    Some of my strongest memories of anxiety come from High School, when I was on the track team, on race days prior to races. My anxiety here didn’t seem to come from a lack of security, since the risk of injury in a race was low. My best explanation for the anxiety is that it was caused by an unknown (and unknowable) outcome in a future situation where there was a reasonable chance there could be a negative outcome (not winning a race). Even towards the end of my senior year when I was typically winning races, I would still be anxious, even though the likelihood of negative outcome was low. As a result, it didn’t feel like a situation I would need to take “flight” from. It also didn’t feel like I was doing any kind of “developing” during the races (hence the ZPD idea doesn’t seem to strongly fit), since race outcomes were mostly determined by the amount of physical conditioning undertaken in the months/years prior to the race. If anything, it seemed like a kind of performance anxiety (will I do as well as I should today?)

    Anxiety before taking a test feels like a similar situation. No physical damage is at stake, and so fight-or-flight circuits don’t seem like they’d come to play. Even when a topic is well known, there is still some anxiety, indicating that here too there is a future situation with unknown outcome (what grade I’ll get) with some risk of negative outcome (questions could be unfairly hard, and lead to low grade). Again, will I do as well as I know I can on the test?

  2. 2 Ian Horswill
    September 19, 2007 at 1:40 am

    Hi Jim,

    Sorry for the delay in replying. I was traveling and got behind.

    I think we actually agree on this. By security here I mean something along the lines of your assessment of the inverse probability of negative event (known or unknown) happening. And anxiety is inverse security, i.e. the probability of an adverse effect happening. Cues to anxiety would then be things like the unknown, the subjective sense that you don’t have control over a situation, or the loss of part of your support structure. I don’t mean to imply that anxiety only happens in the context of development (nor do the attachment theory people claim that). The RST theory does arguably claim that anxiety == goal conflict, be it between the the BAS and the FFFS or within one of those subsystems.

  3. 3 supervalentthought
    January 1, 2008 at 6:56 pm

    Hi, Ian! You should come visit my blog at http://supervalentthought.wordpress.com/. I’ve linked to you there.

    Anyway, I’d like to point out that the two of you use the genre of the *situation* to describe the setting for anxiety. I think this is correct, but since I also guess that you’re using it vernacularly, I thought I’d put my two cents in about why I think it’s worthy to theorize the situation. I quote from a recent paper of mine:

    I call the moment of the historical present a situation deliberately. As we know from situation comedy, a situation is a genre of living that one knows one’s in but that one has to find out about, a circumstance embedded in life but not in one’s control: we have a situation here. When a situation unfolds, people try to maintain themselves in it until they figure out how to adjust. What’s crucial in the management of a situation is the problem of living in the ongoing now of it, the enduring present that is at once overpresent and enigmatic, and which requires getting one’s bearings in new manners of being in it. That’s what makes it an impasse.

    So why I think situation is a major genre for situating anxiety, is that it’s a scene with an impact in the middle of becoming an event. You become split, a subject of the event in process and a person experiencing not knowing the shape of its outcome. Anxiety has a lot to do with ends, don’t you think? So I agree with Ian that you two don’t really disagree.

  4. October 26, 2008 at 3:25 pm

    scientific and useful explanations of anxiety

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: