The Feedback Loop Lens: How Systems Thinking Transforms Habits, Teams, and Well-Being

Introduction: How an AC Explains More Than You Think

It’s peak summer. The temperature outside is 37°C. You’ve just switched on the AC and set it to cool to 24°C.

A sensor detects the external temperature and notes it to be above the target temperature. The controller receives observes this gap and commands the system (in this case the cooling fan) to start the cooling effect. The fan begins cooling, initially with greater intensity due to the high error signal. A sensor constantly checks the room temperature. Gradually, the error signal shrinks, the fan slows, and eventually runs at the speed needed to maintain the target temperature.

This post isn’t about how ACs cool a room; it’s about how most complex systems — teams, habits, cravings, cultures, and many other things — behave. We often think the world runs on linear relations and straight causal links. Instead, it runs on loops.

This generalised flow:

Goal → Sensing → Action → Feedback → Adjustment → Sensing → Repeat

governs not just engineering systems but can also be applied to diverse areas as trusts in early-stage teams, health and habit adoption, food and impulse control, long-term thinking and financial stability etc. Understand the underlying mechanics and knobs, and you’ll begin to see how to use them for your benefit.

Some Definitions and Control Systems Context

A generalised block diagram (via wikipedia)

Before we dive into the human loops, here are a few terms I’ll use throughout:

• Sensor: Detects current system state (e.g., thermometer reading room temperature).

• Error Signal: The gap between target state (24°C) and sensed state (37°C).

• Controller: The logic deciding how strongly to respond. Its sensitivity is called controller gain.

• Noise: Unwanted variability or irrelevant signals that distort the true reading (e.g., a sensor jumping between 26°C and 28°C when the temperature is actually 26.6°C. This impacts the controller.)

(For a deeper dive, jump to the Addendum at the end.)

Loop 1: Trust & Micro-Management

At early-stage startups, culture is still fluid. In fact, even within large firms, small teams often form cultural islands driven by their managers.

It’s quite fashionable to talk about ‘high-ownership’ culture at most organisations. A high-ownership culture necessarily requires a high-trust culture.

“I want to win and my team isn’t best equipped for that; let me hand-hold them through the exact steps needed to achieve this,” believes the manager. They come at it with good intent — a minor deviation from the specification may delay the launch, or ignoring a minor problem now may have far bigger consequences later. Ironically, the manager’s attempts to ‘help’ often erode the trust they need to win in the long term.

The manager notices a gap between the goal (performance) and the signal (team behaviour). They intervene. Hard. Hands-on. Structured. Constant feedback. Clear instructions. Frequent check-ins. From the manager’s perspective, they’re reducing the error.

But from the team’s side? The signal being received might be:

• “I’m not trusted to figure it out.”

• “Every move is being monitored.”

• “There’s no room to experiment or fail.”

Which creates a trust gap, not a performance gain.

What we now have is a feedback loop spiralling in the wrong direction. In control systems language:

• The controller gain is too high: the manager is overreacting to early deviations.

• The sensor may be distorted: the signal — team’s performance — is not an objectively measureable one; thus, the manager may look at other signals (like how vocal they are in meetings or how quickly they start delivering) as a proxy for team’s performance)

• There’s no damping: corrective actions are immediate and intense

• The system lacks integral memory: it lacks the ability to accumulate slow, steady progress and instead reacts to the last ‘snapshot’.

As it stands now, it’s a vicious cycle where the trust gap keeps getting reinforced. Tuning even one of these parameters — slowing the gain, correcting for signal noise, allowing lag — can stabilize the loop. But tune them all, and you create a self-reinforcing trust-building system.

Through first-hand, second-hand (and even third-hand!) accounts I’ve been exposed to a fair number of organisations and their internal cultures. I can confidently say that the wife, through her brief stint at her last job at USAID India, was in an environment of very high-trust. Possibly among the most ‘high-trust environments’ I’ve seen in close quarters.

Between the superiors in her team and her, there was enough freedom to make mistakes and receive timely interventions, such that it neither veered into micromanaging nor directly put one on the deep end of the ocean.

Loop 2: Health, Habits & Virtuous Cycles

Your body — or Instagram — has been giving some signs that you are now inspired to take up a fitness regimen. You don’t trust your willpower and have a history of dropping similar New Year’s resolutions by 14th January. So, you go down the minimal commitment route: it’s better to do it yourself at home. “I could end up quitting within a week, or… (shudder) I could make this sustainable!” you daydream.

But this time, you continue for 3 days in week 1. You’re not discouraged by doing it for only 3 days instead of the targeted 5. You continue onto week 2 as well, again remaining at 3 days. But you notice you’re able to do more repetitions of the same exercises. Soon, you’re at week 3, and you actually look forward to your exercise sessions in the morning. Or to paraphrase a famous quote around writing, you still don’t enjoy exercising, but you like having just exercised.

You also start noticing being more toned, as well as certain bodily signals (improved mood or higher motivation) that are driving you towards exercising regularly. Even if you miss out once in a while, you quickly restart.

That’s a self-reinforcing virtuous loop in action.

Here too: your goal is consistent workouts. Your sensor is how your body feels. The controller might be your mood, willpower, or belief systems — the thing that makes you go or skip today. Over time, as the feedback improves (you feel better), the loop strengthens.

The impact on the sensor is somewhat damped. You don’t see the strengthened muscles immediately after the workout. It takes a few weeks/months before you start noticing anything.

Loop 3: Cravings, Guilt & Vicious Spirals

Conversely, you recall last year. You’d eat an entire packet of chips and feel hungry about 45 minutes later. “No one can eat just one,” after all. Junk or processed food, often, may have that impact on us. We don’t feel full and have more. It causes a certain kind of guilt or induces lethargy.

It's rarely a one-off. The loop kicks in: you eat poorly, feel guilty or drained, which makes you less motivated to cook or move, so you repeat the quick-fix choice. The loop reinforces itself:

Low Energy → Quick Hit → Guilt or Lethargy → Lower Effort → Repeat.

In control systems language, this is a classic positive feedback loop — but one that spirals downwards. The signal (hunger or energy) reaches the sensor (your short-term brain) and triggers the controller (your craving or stress), which then triggers an action (grabbing something that gives an immediate rush) that temporarily masks the problem. But, shortly after, instead of reducing the hunger, it actually amplifies it. You don’t just feel hungry; you crave more of such food. Bad sleep the night before can further distort this signal.

But, compared with the virtuous loop earlier, this time the feedback signal (in the form of craving, dopamine, etc.) is immediate. Instead of the damping effect (when the error signal gradually reduces) there’s either an amplification effect at play or a very low damping effect.

Breaking Bad Loops

Due to the looping nature, it’s not sufficient to ‘break the cycle’ once. Their self-reinforcing nature means that even if you skip it once, the loop will resume.

I’ve on and off struggled with quitting cigarettes. I’d quit cold turkey and was good for a few months. Then, I randomly smoked a cigarette. And before I knew, three more. The last smoked cigarette incites a craving soon after, and my short-term brain acts on that craving immediately. No time-delay between the action (last smoked cigarette) and the feedback signal (craving for the next) and a signal to the controller to act on this craving yet again.

So, what do we do? How do we break the cycle?

In the context of addictions, it’s not a willpower game. We’re wired to respond to what’s immediate, not what’s meaningful. Our sensors are noisy; our gain is too high; and our memories are short.

To escape the spiral, we don’t need brute force — we need better tuning. Dial down the controller gain so every craving doesn’t demand a reaction. Add time lag between impulse and action, so the system has a chance to settle. And clean the signal: learn to distinguish true hunger (there’s no ‘true’ desire to smoke) from noise (stress, boredom, habit).

These are some means of injecting a stabilising cycle into a wobbly loop.

Building Virtuous Ones

Building virtuous loops, or as referred to in startup-speak — building sustainable flywheels — helps if the positive reward feels immediate. A tasty post-workout smoothie that you drink only on the days of the workouts, could be one example.

Eliminating freedom and discretion, and implementing things ‘as a rule,’ helps. Every time you ask if you should work out today or postpone to tomorrow, you leave it in the hands of the controller (mood), which may make us skip that dreaded leg day. Instead, implementing a blanket rule (such as “I’d exercise between 0730 and 0830 hrs on Mon, Wed, Fri every week”) reduces the impact of noise on the controller.

Or, breaking the streak once isn’t a signal to go off the rails. Building ‘failure’ into the system as a non-failure (“I won’t miss twice in a row,” or making it easier to restart after a gap) helps too.

And, So? The Unseen Loops Around Us

The true power of understanding control systems extends far beyond ACs or engineering. Most of us naturally think in simple cause-and-effect or linear chains of events, but this framework offers a powerful lens to view the world through feedback loops. These feedback-driven systems explain why things spiral — up or down — and how small interventions in gain, delay, or signal quality can change everything.

This understanding also illuminates concepts like how the signals we project can shape our personality through the feedback we receive, a topic I explored in a previous post.

This was a primer. In the next post, I intend to expand on the tunable parameters (gain, lag, noise) using a few more examples (poverty, habit design), and hopefully also covering what’s called as second-order thinking or ‘systems’ thinking.

Addendum: Control Systems for the Curious

Control Systems was among the select few subjects in my engineering that I found interesting. It broke down complex, messy systems into simpler basic blocks: sensors, controllers, and systems. One needn’t care about what happened inside these individual blocks. It then mathematically modelled how these blocks changed the input signals into output signals. But the real fun was with the systems that had a feedback mechanism, where the output of a loop would turn into the system’s input in the next turn.

Generalised block diagram again (via wikipedia)

This is a classic negative feedback loop in action. When analysing a given feedback loop, components to keep in mind:

• Quality of Sensor Output: What is the nature of the output signal of the sensor? Is it a measurable and quantifiable number (temperature verus team performance)? Does it generate false alarms? Does it have low noise (i.e., is the output signal accurate and narrow, not fluctuating or carrying ambiguous information)?

• Logic of Controller: Controller is the brain that instructs the system. How does it convert the error signal into the instruction that the system can understand? Does it have a damping effect (gradually reducing the error to zero)? Does it reduce it to zero aggressively or with gentle tuning? As a result of being aggressive, does the error signal oscillate around zero?

  • This is represented by ‘controller gain’ — a high controller gain means a more aggressive ‘chasing’ of the target. e.g., an AC fan going full blast mode given even a 1°C temperature increase is ‘high gain’. It corrects too much, too fast.

• System Response: How quickly does the system respond to the input from the controller? How soon after the system’s response does the output begin changing?

Two other variants to note:

• Positive Feedback Loop: Instead of the error being minimised, it is amplified (e.g., a mic amplifying its own output; a social media outrage cycle caused by a faux pas and amplified by doubling down)

• Feedforward Loop: Here an action is taken based on expectation instead of actual feedback (cleaning the house before your wife returns some guests visit; preparing the meals or medicines for the week ahead).