Enhance Decision-Making With Control Systems

This article is part of our exclusive career advice series in partnership with the IEEE Technology and Engineering Management Society.

Much of engineering is decision-making. Engineers make decisions about product design, program management, technology road maps, research directions, leadership of technical teams, and more.

As a past president of the IEEE Control Systems Society and now the 2026 president-elect of the IEEE Technology and Engineering Management Society, as well as holding leadership positions in industry and academia, I have thought a lot about the connections between control systems and technology management.

The safe, reliable performance of airplanes and spacecraft, cars and trucks, homes and buildings, chemical plants and manufacturing facilities, communication and financial networks, and many other complex systems relies on automation and control systems. But, as I discuss here, the concepts of control engineering are also relevant to human decision-making in technology management.

Whether in engineering or management, uncertainties are pervasive. In the case of the latter domain, we can never be sure about innovation processes, market projections, and people’s personalities and capabilities. Indeed, the uncertainties may seem so overwhelming that some might be tempted to make a decision by flipping a coin.

But most decisions are not made randomly, and control engineering offers insights for managerial decision-making under uncertainty.

Mental models and uncertainty

We rely on mental models—our knowledge, beliefs, assumptions, experiences, observations, and reasoning. But models of any variety are not reality. They are accurate approximations at best, and they’re completely wrong at worst. It is essential that all decision-makers recognize the discrepancies between their mental models and reality, and then take action to reduce the mismatch.

Let me draw an analogy from control engineering. To develop a control system for an aircraft, for example, mathematical models—not the mental variety—are developed of the plane’s airframe. For numerical accuracy, the models require “sufficient excitation,” which means providing a variety of inputs, such as deflections of flight control surfaces, and measuring how the airplane reacts to them.

Based on that data, models of the required accuracy can be created and incorporated into the flight controller design. The data must be rich enough so that relevant signals can rise above irrelevant noise.

Decisions are rarely one-and-done affairs. Leading a team, managing a project, allocating resources, and undertaking a design all require regular interactions with others, with initial decisions adjusted regularly over time.

The same applies to mental models for human decision-making. Monitoring normal day-to-day operations of an organization or a project likely would not provide information of a high enough signal-to-noise ratio for mental models to be reliably updated.

Instead, special tasks and situations can be instrumental in achieving the goal. For example, a manager could give a challenging task to a team member primarily to improve the manager’s mental model of the employee, rather than to address a pressing organizational need. The improved mental model can help the leader determine the best role for the employee when an actual challenging situation arises.

Regardless of effort, mental models will never be perfect. There will always be uncertainty. So, one crucial lesson for decision-makers to keep in mind is that whatever you know, you only think you know. Resist the temptation to believe you really know the truth.

As a decision-maker, the objects of your mental models include your organization, other stakeholders, and the external environment. But they also include your self-model. You need to have a clear understanding of your own capabilities, preferences, and circumstances. Examples include your workload, the pace at which you work best, your flexibility in light of other priorities, and what motivates you. And, of course, you need to appreciate that your self-models are uncertain, too.

People often don’t know themselves as well as they think they do. Be honest with yourself, and ask for feedback from trusted colleagues and friends. Don’t react defensively; listen to the feedback, then reflect. Doing so can strengthen your understanding of yourself.

Dynamics and decision-making

Sometimes the effects of a decision aren’t immediately apparent. It can take days or even years for that to happen. In the meantime, observations can provide an indication of the effects, but they could also be wrong. In control theory, for example, we teach the concept of inverse response, where the initial response to a decision is the opposite of the final effect.

A simple example is what happens to a company’s profits if it significantly increases its research and development investment. For the next few quarters, profits likely will be lower because of the R&D expenses. Once new products roll out, profitability probably will increase.

A manager who doesn’t recognize the temporary inverse response trend and cuts R&D resources can worsen rather than improve matters by sacrificing the long-term vitality of the company. Such short-sighted decisions happen all too often.

Decisions are rarely one-and-done affairs. Leading a team, managing a project, allocating resources, and undertaking a design all require regular interactions with others, with initial decisions adjusted regularly over time.

Those dynamics must be considered in complex decision-making situations. The adjustments are based on monitoring the activity, thereby closing the feedback loop.

Time delays can be especially difficult to manage. As noted, decisions made about projects and processes take time to have an impact. Delays can result from various sources including communication issues, new policies, staffing problems, procurement times, and reporting processes.

To be an effective decision-maker, your mental model should include estimates of delays. The complications arising from unanticipated setbacks in feedback processes are well known, both in control engineering and systems engineering. The ability to anticipate delays—and, where possible, to reduce them—is a valuable skill for decision-makers.

Connecting the dots

The interconnections among the concepts of mental models, uncertainty, dynamics, and feedback are deep and fascinating. The insights they offer for decision-making are numerous.

One example is the robustness-performance tradeoff in control engineering. The tradeoff refers to the fact that the highest levels of performance cannot be attained while simultaneously being robust during times of high uncertainty. This insight is the basis of the “no free lunch” theorem in optimization, meaning that no decision-making approach can be optimal in all situations.

When uncertainty levels increase from a mismatch between a mental model and reality, the presence of noisy data, or external disturbances, decision-making should be less aggressive. Instead, you should respond by making gradual changes and waiting for feedback signals. To paraphrase, the more uncertain the situation, the more one should hedge one’s bets.