Resilience (2 of 5): The feedback dynamics between you and academia
Dr Emily Troscianko and Dr Rachel Bray
In the first post in this series on resilience, we thought about how you and the academic environment interact. We considered some of the benefits and drawbacks of working in academia, and some of the personal characteristics that might make you more or less resilient in the face of its stresses.
In this post, we offer a way of modelling these interactions in more visual, dynamic terms. Seeing things drawn as well as written can be helpful in understanding the underlying structures (it’s not just about me…), and in adopting a fresh perspective on choices available to you.
Once you start thinking of yourself, your life, and the world as a feedback system, it can become an addictively satisfying way of seeing the bigger picture. Feedback dynamics are fundamental to pretty much any system you can think of – biological, technological, institutional, psychological… This means that grasping some feedback basics is crucial to understanding how instability arises within and between the systems that make you you – and equally crucial to working out how a happier stability can be reinstated.
Let’s start by looking at what happens in the case of a system where there is no feedback, i.e. an open-loop system. Imagine a car:
The fuel is an input into the system which is used to control the output, the car’s speed. If there were no other factors to worry about, we could confidently assume that x amount of fuel would result in x mph. But this is a car in an environment: there are things like wind and gradient which need taking into account, and which are beyond our control; we can call them disturbances (or in the technical context, perturbations).
Whether we’re thinking about a car or your career, disturbances are guaranteed. This means we need to measure the output and check whether it’s really what we think it is, and if not, make some adjustments. This is feedback control:
Here you have an aim in mind – the speed you want the car to be going – and you design a controller to make that happen. The cruise control measures the difference between the actual speed and the desired speed and uses that value (the error) to adjust the fuel and therefore the car’s speed. And if the controller is designed well, then things like wind speed and road gradient (the external disturbances) can change but the speed will still remain near-enough 70 mph. Then we say that the system is robust. Robustness, in feedback engineering, is roughly equivalent to what in a psychological system we might more colloquially call resilience.
So, let’s move on to a system that encapsulates some of what we might care about in the academic context:
Let’s imagine that what you’re aiming for is something like a balanced and fulfilled life (the set point), and the controllers are major work/life sub-systems like job and family. They collectively determine factors like the amount of time you spend doing different things, which have effects on variables like your happiness and your professional effectiveness. When an error is detected between the reality of those variables and the state you’re aiming for – that balanced life – an adjustment is made in your job and family (or whatever other components are important in your life). Making an adjustment could involve altering the hours spent on each component, as well as effort invested, material resources acquired, social resources drawn on, and many less easily measurable variables.
In this system, disturbances beyond your control might be anything from having to take up the slack for an absent colleague to coping with a deadline or a family emergency. The test of the system’s resilience is how well it maintains something like a balanced life despite those disturbances. And that will depend on factors like how timely your responses are, and how well the controller(s) learn from past responses which ones are most likely to work best given a particular disturbance or error type. All these characteristics are in turn shaped by a starting state of the system: everything that’s defined by prior conditions like your personality traits, childhood upbringing, and socioeconomic conditions.
When a disturbance intrudes on the system, there are two ways for things to go: into a spiral of self-propagating instability, or towards a reinstatement of stability.
Consider these two possibilities for your life as an early-career academic. Say you’re starting at the enviable point of a balanced life, and then something changes, and a chain reaction gets started, maybe something like this:
Balanced life → colleague falls ill → work hours increase → work quality decreases → work hours increase → family time decreases → happiness decreases → life less balanced → work hours increase to meet targets → happiness decreases further → etc. etc.
One can imagine that another disturbance like, say, a grant rejection somewhere in this cycle would increase the instability even further. But stabilisation is always possible. Maybe in as straightforward a way as this:
Balanced life → colleague falls ill → work hours increase → work quality decreases → work hours increase → family time decreases → happiness decreases → life less balanced → partner notices decrease in happiness and makes you take a weekend completely away from work → life more balanced → happiness increases → work quality increases → etc. etc.
This example reminds us of a couple of other things too. First, nothing about you happens in isolation from anything around you: the system is ‘You’, yes, but also Job, Family, or whatever else comes in the controller box for you. Second, our own successes and failures of resilience affect the other systems that involve other people: if your partner also takes up the slack for you by doing more than their fair share of household tasks, or has to cope with your lowered mood often or for extended periods, their resilience will be reduced and the likelihood of unstable, self-reinforcing feedback will also grow.
Every feedback diagram we might draw is always nested within infinite others which expand to represent the dynamics of the whole universe. We represent feedback dynamics with a resolution that’s relevant to the context, and we could always zoom in or zoom out.
For our purposes here, the feedback system that is the university interacts with the one in Figure 3 in obviously important ways. The university’s set points might be financial and reputational; its controller might be the Vice-Chancellor; some of its inputs might be students, staff, and income; some outcomes might be degrees, jobs, and expenditure; and so on. We can envisage the interlocking of this and your system along these lines:
Any of the university’s functions can act as inputs to the ‘personal’ system, whether in terms of how university priorities affect one’s working habits, or admissions affect research/teaching productivity, or recruitment patterns affect job/grant success, or any number of other iterations.
Feedback dynamics can span the personal/institutional divide in both stability-enhancing and stability-reducing ways. An example of self-correcting, stability-enhancing feedback might be when recognition of and protest at collectively unsustainable working habits prompts an institutional review and an attempt at greater or changed regulation. An example of self-reinforcing, unstable feedback would be when your perfectionism and anxiety disrupt your teaching and research activity, generating institutional pressures on you which you’re unable to meet, resulting in a breakdown. (See Troscianko 2017 for an example of how feedback loops are manifested in mental and physical illness.) Here, stability is ultimately restored, but only by a radical adjustment like your withdrawal from the exacerbating context.
The capacity to turn potentially unstable dynamics back into stable ones will depend a lot on you: on self-insight, on the willingness to change and the knowledge that change can’t be infinite. A good way to begin is to be open-eyed towards the things you can’t change, to make your peace with them, and then to identify the factors you can change and address these pro-actively. Sometimes a large change is needed to restore your equilibrium – moving closer to where you work, say – and sometimes an apparently tiny thing is enough: logging out of your email more often while you work.
Take a moment to reflect on the dynamics of action and reaction in your life right now. Maybe even sketch a feedback loop or two to help you! You can then base your perseverance within the university system on genuine acceptance of its realities, and make better-informed decisions about when and how to persevere within it, and when and why to call a halt to the perseverance, remembering that there are other options out there.
A technical appendix:
The two types of feedback loop we’ve considered, the one tending towards instability and the one tending towards stability, have somewhat counter-intuitive technical names. They are positive feedback and negative feedback.
In English, of course, positive has two meanings:
1) good, or beneficial
2) greater than zero
Likewise, negative means:
1) bad, detrimental
2) less than zero
Especially when coupled with the word feedback, the evaluative meanings often trump the numerical ones. For our purposes here, this is unfortunate, since positive and negative feedback are great ways of conceptualising self-perpetuating and self-cancelling feedback dynamics. A classic example of positive feedback is the screech of a microphone picking up the sound of a speaker being used to amplify it: the microphone feeds the speaker sound back to the speaker to be amplified, then the speaker amplifies that sound and feeds it into the microphone, and so on until your ears hurt. This is positive feedback because it’s additive: each iteration adds to the last in a rising spiral. Negative feedback, by contrast, is self-cancelling, tending to damp itself down. A familiar example is the cruise control we considered at the start, which measures the actual speed, subtracts it from the desired speed, and uses this signal to compute how much more or less fuel to inject, resulting in a new speed to measure. If the cruise control is working well, the fluctuations should get smaller and smaller, rather than bigger and bigger, until a neat 70 mph is reached – and then adjust quickly again if you suddenly start down a steep hill.
If you want to read more on positive and negative feedback in the systems engineering not the TripAdvisor sense, you can do worse than start with Wikipedia, here and here.
With thanks to James Anderson for the feedback engineering advice. James completed his DPhil in Engineering at Oxford in 2012 and is now a senior postdoc at Caltech.
In the first post in this series on resilience, we thought about how you and the academic environment interact. We considered some of the benefits and drawbacks of working in academia, and some of the personal characteristics that might make you more or less resilient in the face of its stresses.
In this post, we offer a way of modelling these interactions in more visual, dynamic terms. Seeing things drawn as well as written can be helpful in understanding the underlying structures (it’s not just about me…), and in adopting a fresh perspective on choices available to you.
Once you start thinking of yourself, your life, and the world as a feedback system, it can become an addictively satisfying way of seeing the bigger picture. Feedback dynamics are fundamental to pretty much any system you can think of – biological, technological, institutional, psychological… This means that grasping some feedback basics is crucial to understanding how instability arises within and between the systems that make you you – and equally crucial to working out how a happier stability can be reinstated.
Let’s start by looking at what happens in the case of a system where there is no feedback, i.e. an open-loop system. Imagine a car:
 |
| Figure 1 |
The fuel is an input into the system which is used to control the output, the car’s speed. If there were no other factors to worry about, we could confidently assume that x amount of fuel would result in x mph. But this is a car in an environment: there are things like wind and gradient which need taking into account, and which are beyond our control; we can call them disturbances (or in the technical context, perturbations).
Whether we’re thinking about a car or your career, disturbances are guaranteed. This means we need to measure the output and check whether it’s really what we think it is, and if not, make some adjustments. This is feedback control:
 |
| Figure 2 |
Here you have an aim in mind – the speed you want the car to be going – and you design a controller to make that happen. The cruise control measures the difference between the actual speed and the desired speed and uses that value (the error) to adjust the fuel and therefore the car’s speed. And if the controller is designed well, then things like wind speed and road gradient (the external disturbances) can change but the speed will still remain near-enough 70 mph. Then we say that the system is robust. Robustness, in feedback engineering, is roughly equivalent to what in a psychological system we might more colloquially call resilience.
So, let’s move on to a system that encapsulates some of what we might care about in the academic context:
 |
| Figure 3 |
Let’s imagine that what you’re aiming for is something like a balanced and fulfilled life (the set point), and the controllers are major work/life sub-systems like job and family. They collectively determine factors like the amount of time you spend doing different things, which have effects on variables like your happiness and your professional effectiveness. When an error is detected between the reality of those variables and the state you’re aiming for – that balanced life – an adjustment is made in your job and family (or whatever other components are important in your life). Making an adjustment could involve altering the hours spent on each component, as well as effort invested, material resources acquired, social resources drawn on, and many less easily measurable variables.
In this system, disturbances beyond your control might be anything from having to take up the slack for an absent colleague to coping with a deadline or a family emergency. The test of the system’s resilience is how well it maintains something like a balanced life despite those disturbances. And that will depend on factors like how timely your responses are, and how well the controller(s) learn from past responses which ones are most likely to work best given a particular disturbance or error type. All these characteristics are in turn shaped by a starting state of the system: everything that’s defined by prior conditions like your personality traits, childhood upbringing, and socioeconomic conditions.
When a disturbance intrudes on the system, there are two ways for things to go: into a spiral of self-propagating instability, or towards a reinstatement of stability.
Consider these two possibilities for your life as an early-career academic. Say you’re starting at the enviable point of a balanced life, and then something changes, and a chain reaction gets started, maybe something like this:
Balanced life → colleague falls ill → work hours increase → work quality decreases → work hours increase → family time decreases → happiness decreases → life less balanced → work hours increase to meet targets → happiness decreases further → etc. etc.
One can imagine that another disturbance like, say, a grant rejection somewhere in this cycle would increase the instability even further. But stabilisation is always possible. Maybe in as straightforward a way as this:
Balanced life → colleague falls ill → work hours increase → work quality decreases → work hours increase → family time decreases → happiness decreases → life less balanced → partner notices decrease in happiness and makes you take a weekend completely away from work → life more balanced → happiness increases → work quality increases → etc. etc.
This example reminds us of a couple of other things too. First, nothing about you happens in isolation from anything around you: the system is ‘You’, yes, but also Job, Family, or whatever else comes in the controller box for you. Second, our own successes and failures of resilience affect the other systems that involve other people: if your partner also takes up the slack for you by doing more than their fair share of household tasks, or has to cope with your lowered mood often or for extended periods, their resilience will be reduced and the likelihood of unstable, self-reinforcing feedback will also grow.
Every feedback diagram we might draw is always nested within infinite others which expand to represent the dynamics of the whole universe. We represent feedback dynamics with a resolution that’s relevant to the context, and we could always zoom in or zoom out.
For our purposes here, the feedback system that is the university interacts with the one in Figure 3 in obviously important ways. The university’s set points might be financial and reputational; its controller might be the Vice-Chancellor; some of its inputs might be students, staff, and income; some outcomes might be degrees, jobs, and expenditure; and so on. We can envisage the interlocking of this and your system along these lines:
 |
| Figure 4 |
Any of the university’s functions can act as inputs to the ‘personal’ system, whether in terms of how university priorities affect one’s working habits, or admissions affect research/teaching productivity, or recruitment patterns affect job/grant success, or any number of other iterations.
Feedback dynamics can span the personal/institutional divide in both stability-enhancing and stability-reducing ways. An example of self-correcting, stability-enhancing feedback might be when recognition of and protest at collectively unsustainable working habits prompts an institutional review and an attempt at greater or changed regulation. An example of self-reinforcing, unstable feedback would be when your perfectionism and anxiety disrupt your teaching and research activity, generating institutional pressures on you which you’re unable to meet, resulting in a breakdown. (See Troscianko 2017 for an example of how feedback loops are manifested in mental and physical illness.) Here, stability is ultimately restored, but only by a radical adjustment like your withdrawal from the exacerbating context.
The capacity to turn potentially unstable dynamics back into stable ones will depend a lot on you: on self-insight, on the willingness to change and the knowledge that change can’t be infinite. A good way to begin is to be open-eyed towards the things you can’t change, to make your peace with them, and then to identify the factors you can change and address these pro-actively. Sometimes a large change is needed to restore your equilibrium – moving closer to where you work, say – and sometimes an apparently tiny thing is enough: logging out of your email more often while you work.
Take a moment to reflect on the dynamics of action and reaction in your life right now. Maybe even sketch a feedback loop or two to help you! You can then base your perseverance within the university system on genuine acceptance of its realities, and make better-informed decisions about when and how to persevere within it, and when and why to call a halt to the perseverance, remembering that there are other options out there.
A technical appendix:
The two types of feedback loop we’ve considered, the one tending towards instability and the one tending towards stability, have somewhat counter-intuitive technical names. They are positive feedback and negative feedback.
In English, of course, positive has two meanings:
1) good, or beneficial
2) greater than zero
Likewise, negative means:
1) bad, detrimental
2) less than zero
Especially when coupled with the word feedback, the evaluative meanings often trump the numerical ones. For our purposes here, this is unfortunate, since positive and negative feedback are great ways of conceptualising self-perpetuating and self-cancelling feedback dynamics. A classic example of positive feedback is the screech of a microphone picking up the sound of a speaker being used to amplify it: the microphone feeds the speaker sound back to the speaker to be amplified, then the speaker amplifies that sound and feeds it into the microphone, and so on until your ears hurt. This is positive feedback because it’s additive: each iteration adds to the last in a rising spiral. Negative feedback, by contrast, is self-cancelling, tending to damp itself down. A familiar example is the cruise control we considered at the start, which measures the actual speed, subtracts it from the desired speed, and uses this signal to compute how much more or less fuel to inject, resulting in a new speed to measure. If the cruise control is working well, the fluctuations should get smaller and smaller, rather than bigger and bigger, until a neat 70 mph is reached – and then adjust quickly again if you suddenly start down a steep hill.
If you want to read more on positive and negative feedback in the systems engineering not the TripAdvisor sense, you can do worse than start with Wikipedia, here and here.
With thanks to James Anderson for the feedback engineering advice. James completed his DPhil in Engineering at Oxford in 2012 and is now a senior postdoc at Caltech.
Links to all posts in this series:
- Does academic complement or conflict with who you are?
- The feedback dynamics between you and academia [this post]
- Distinguishing between assumptions and reality [next post]
- Your working identities
- Having a strategy for cultivating resilience
For unabridged versions of these posts and full references, check out the Resilience Hub.