Effective Communication by Manufacturing Empathy

Posted on March 4, 2017 by Gabe Parmer

Communication with other humans is difficult. Many interactions in academia revolve around either attempting to convey information to others, convincing others of your ideas, learning from others, or critically examining other’s ideas. All of these situations involve information and experience asymmetries – some individuals know more about an area or idea than others. The goal of communication is to increase the total knowledge of all individuals involved. When trying to convince others of your ideas, the goal is also to raise their knowledge to a level where they understand and agree with the value of your approach. This can be quite difficult, and contemplating how to do it effectively is necessary for researchers.

Why is Communication Essential for Research?

First, lets detour and discuss why we care about this. Academia often requires groups of researchers, all attacking a set of shared problems of interest, all approaching the problem from different perspectives. Why is this? The total context, knowledge, and experience of a group of individuals is inherently larger than that of an individual. Though this seems like a sufficient justification for researching in groups, I’ve seen many skeptical students. This skepticism comes from the fact that doing research requires a deep-dive into a topic, and the perception can be that once you have reached a sufficient depth, group interactions lose value. This perspective is dangerously ignorant of two facts:

1. Any research area requires knowledge in the intersection of a large number of other sub-areas, creating a large venn diagram of research areas. Even if you’re an expert in a few of these areas, others will teach you in their areas that will enhance your understanding in your original area. Ideas are not independent, and exist in an eco-system. Having a group that reflects that eco-system is necessary if we are to do strong research.
2. Creativity is required for research, and pushing only into the depths of a small set of areas removes one of the most powerful creative tools we have as humans: the ability to change the context we use to look at an idea. If you approach an area from only a single perspective, the set of ideas you’ll generate will start to stagnate over time. New perspectives from other subjects enables your mind to refocus on the problems you’re solving with new context which is a fundamental driver of creativity. Since research in systems is often the product of applying creativity to engineering, and studying it scientifically, you cannot afford to hinder your own creativity.

I hope that this debunks the idea of the effectiveness of the “lone wolf” researcher. At the very least, it should motivate a genuine commitment to being an active member in the broader research group. To foster a group environment where each member can benefit from these factors, see the post on psychological safety.

How to Effectively Communicate

It is deceptively simple to think that because we talk to other humans all day long, and have been doing so for most of our lives, we can effectively communicate in an academic environment. This is generally incorrect. I think it is important to first acknowledge that effective communication is really difficult, especially in deeply technical topics with large information asymmetries. Second, remember that these discussions should be relatively ego-less, thus the goal is not demonstrating knowledge or intelligence, rather growing the level of collective knowledge. In my experience, if someone perceives the goal of interactions as an opportunity to demonstrate knowledge, it ends up being a large waste of everyone’s time.

First, lets focus on some individual factors that we can each to be increase our effectiveness of communicating.

• Understand explicitly What is your goal for the interaction? If you have a meeting scheduled with someone, what is the point of it? What do you want the other member of the meeting to learn, or what do you want to learn from them? Without a set of goals, it is unlikely that a technical discussion will yield much useful information transfer. It is important for your goals to match those of the other parties; always make sure that meetings have a stated purpose.
• What do you need to prepare given your goals? What do you need to learn or organize to aid in achieving those goals? Especially if you’re trying to learn about a topic from someone else, doing some initial investigations will enhance your ability to learn, and utilize their time more effectively.

This list of things we can do independent of others to aid in communication is a very small. There isn’t much that you can do to optimize your communication with someone else while only taking yourself and your ideas into account. To most effectively communicate, you need to focus on the other party’s state. To emphasize this fact, a set of examples that show that the requirements on communication vary based on the other party much more than on your or the topic.

Explaining your research. You want to explain your research to

• Your mom.
• Your significant other.
• A high-school student taking their first programming class.
• A CS undergraduate freshman.
• A CS undergraduate senior.
• A CS PhD student in a different field.
• A CS PhD student in a similar field (i.e. in the lab).
• A venture capitalist.
• A CEO.
• A Dean of engineering (not just CS).
• A CS professor in a different area.
• An academic you meet at a conference.
• Your advisor.

Go through the list, and give a three to six sentence summary of the work you’re doing for each audience. If there is significant overlap in your explanation for many of these categories, you’re doing it wrong.

Understand how a subsystem works. You want to understand how a subsystem of a system that you’re unfamiliar with works, and you ask

• A highly technical engineer.
• A project manager.
• A PhD student.
• A professor that is removed from the details.

How do you state your questions? How do you react and followup to their answers? Again, this should likely deviate for each of these groups.

Explaining a research paper. You explain a paper to

• Your mom.
• Your research group.
• Your professor.
• A group of undergraduates.
• A group of your peers at a research conference.

Again, the methods you use should have significant deviation. There are many other examples, but I hope this set makes the point.

These examples demonstrate the pervasive wisdom that you should “know your audience”. This is likely one of the most important aspects of communicating effectively with other people. It takes genuine effort and thought to contemplate what your audience knows, and how to convey your information, or how to state your questions in a manner that is most effective.

Manufacturing Empathy¹

Many people associate empathy with a vague notion of sharing another’s feelings. I like a definition that is little bit more concrete:

Empathy is the capacity to understand or feel what another person is experiencing from within the other being’s frame of reference, i.e., the capacity to place oneself in another’s position.

Manufacturing empathy expands on the trite concept of knowing your audience. It is not a passive activity that occurs before you make a presentation, and instead an active activity that you perform while engaged in a community, or a conversation. The point is to always work toward a greater understanding about the other’s perspectives, knowledge, and understanding. This allows you to much more effectively state your concepts or questions in a manner that is understandable, and always advancing the conversation toward greater collective knowledge.

Manufacturing empathy is a tacit acknowledgment that it takes hard work, dedication, and concentration to effectively state your thoughts in a manner that will resonate with and be impactful on other humans. It is motivated by the fact that it is more important to focus on the other person’s understanding of the information, than on the information itself.

Building mental model of other’s state. One way to describe empathy, is that it is concerned with building descriptive mental models of the other’s state and mind. These mental models have a number of dimensions.

• Goals. What are the goals of the other person? Understanding where the goal of the conversation is to the other person is essential to being able to continuously move the conversation toward that goal. Without a goal, the conversation is aimless and is unlikely to meaningfully increase collective knowledge. It is also important to acknowledge early on if there is a goal disparity between the parties. If there is, that needs to be addressed before move on to other discussion.
• Interests. What are the interests of the other person? To fully engage someone and make them want to understand the points you’re making, you have to understand what they are interested in. Tailoring your explanations and questions toward those interests will maintain a higher level of engagement, thus increasing the chance that you’ll meet your goals.
• Knowledge. What does the other person understand about each of the pieces of required knowledge to convey a point? All information builds on the understanding from an existing base of knowledge. Explicitly building this base of knowledge is called “scaffolding”, and is required in any methodical delivery of complex ideas. Even when learning from another, helping build their scaffolding to understand what you understand is necessary.
• Expertise. What are the areas of expertise of the other person? A person’s areas of expertise are more specific than their general breadth of knowledge or interests, and if it is somewhat close to the area of conversation, it is often one of the best sources of analogies. Analogies are an effective way to explain some concepts, especially to those with very little knowledge in the areas you need them to understand. Analogies must be comfortably in the areas of knowledge of the other person if they are to be effective, and are most effective when they derive from their area of expertise.
• Confidence. What is your confidence in your mental model for each of the above dimensions? If you’re used to talking to someone, your models should be highly accurate. This is the benefit of mentoring models like the PhD student/advisor relationship – an hour meeting can be very productive as a majority of time is spent actually increasing collective knowledge. An initial meeting with an arbitrary high-school student is difficult as any mental models you have are not very reliable. This leads to the “interview” problem: it is very difficult to build a high-confidence, descriptive mental model of someone who is interviewing for a job in the meager time allocated to interacting with them.

So if manufacturing empathy requires an active effort to build and refine your mental models of the other people you’re interacting with, what actions should one take? Techniques for manufacturing empathy include:

• Ask more questions than you convey facts. It is much more important to understand what the other person is trying to convey and understand their state of mind, than to continuously convey knowledge. Most of a discussion should revolve around asking questions. This applies equally to the person that is attempting to convey information to the other, and the person that is naturally asking questions. Asking questions to understand where someone is coming from should always be the highest priority.

• Focus more on how people reply to your comments than on what you’re going to say next. How the other people react to your comments is your main source of input about their mental model. How deeply do they understand the area your having a conversation about? Do they understand the points that you’re trying to make? Many people make a point, hear a reply, and move on to the next point, without much of a reflection on how the reply reflects on the other person’s actual understanding. This is a waste of time for everyone involved. The most common response to someone’s reply to one of my questions, or to a point that was just made is often “what do you mean by that”.

• Be attentive to ambiguity. When the other person states something that is consistent with the understanding you want them to have, spend some effort making sure that there is no ambiguity in their reply. Especially in technical areas, it is not uncommon for a single assertion to have many different interpretations. When someone is building their knowledge-base, their questions and comments are often ill-stated because they don’t have the experience formulate a consistent reply. It is very easy to see the answer you want in an ambiguous statement. However, it is important to understand that ambiguity can hide a lack of understanding. Don’t move on to the next point, and instead discuss the question to build a better mental model.

• Repeat the points the other person is making in your own words. This is, in effect, a way to check the consistency between your model of the person’s understanding, and their actual understanding. When you restate in different terms what the other is trying to say, you’re expressing your model of what you think they know. You’re giving them a chance to debug that model, interactively.

• Slow down. Most people seem to think that flying from point to point is the way to have a productive discussion. It may be the case that you “cover” more material in a given amount of time by essentially enumerating facts. However, it has almost no value as it runs counter to increasing knowledge and understanding. Time has to be paid to making sure that both people’s understanding is consistent. Equally harmful to conveying knowledge and understanding is being distracted by tangents. Tangents have a tendency to make it hard to maintain a strong mental model of where the other person is at because they, by their very nature, only distract from the main goals of the conversation.

All of these mechanisms are methodically building feedback loops into the formation of your mental model of the other person. The other person provides more information to you about their own mental state, and you check this information’s consistency with your model, and refine the model as necessary, often through further interaction. This methodology is counter to a sense of confidence that we can convey complex information easily.

Provide Feedback to Enhance Other’s Empathy

It is the job of each member in a conversation to make sure that each individual achieves their goals. Put another way, your job goes beyond just to saying your piece, or learning what you want. It is important, once you approach your interactions from the perspective of building empathy, to understand that you must enable others to do the same. This is quite a bit easier than building empathy, and largely centers around providing constant feedback on what you believe the state of the conversation is. This takes a number of forms:

• When you don’t understand something, say so, and provide as much information about your confusion as possible. It is very common for researchers to feel like they “should” understand something, so they will act like they do. This is a waste of everyone’s time. It is important to state this feedback carefully and respectfully, as some people can take it as a personal attack that they aren’t explaining their idea well. If you’re talking to me, don’t pull your punches; be as direct as possible.
• When you do understand something, let the other person know, and restate or summarize their point to ensure that you’re both on the “same page”, or that your have a mutual understanding. This is exceedingly important because this is the signal that the conversation can move on to the next point. If you feel like some conversations are much longer than they need to be, and that explanations are drawn out, it is likely because it is ambiguous when you understand, and can move on. Professors are particularly bad at “over-explaining” concepts. This is because 1. we understand that discussion while failing to conveying ideas is a waste of time, and 2. we are used to explaining ideas to people of varying knowledge levels, so usually conservatively explain toward a broader range of knowledge levels.
• Clearly state when you want an elaboration. This differs from not understanding something, in that you have learned something that opens up a new area of interest. Sharing that you’d like to learn more about that is essential for the other person to build a mental model of your interests.
• Repeat your understanding of the other person’s points. Even if you believe that you’re in mutual understanding, it is often important to continuously provide feedback on the state of your own understanding of the other person’s points.

Practice

As with any skill, manufacturing empathy takes practice. Research groups often provide this opportunity in a few different ways:

• scrum and other forms of weekly meetings
• research meetings often centering around paper presentation
• more informal paper discussions
• research/poster days
• practice/mock paper presentations
• lunches with fellow students
• meetings with other students and with your advisor
• teaching and being a teaching assistant

It is important that these initiatives are student-motivated and run so that they have genuine momentum and longevity. Hopefully this provides yet another motivation to organize these events.

Beyond Academia

I’ve framed this discussion in the context of academia. The general technique goes far beyond this context. A number of examples, and I’ll let you expand from there:

• Job interviews. Your job is to always figure out not just what the answer to each question in a technical interview is, but also what the person sitting across from you wants out of the interview. Often answering “correctly” isn’t as important as satisfying the goals of that person.
• Teaching. Many developers continuously have to teach their colleagues, and be taught by them. Teaching is one of the rare opportunities where you not only can practice building empathy, but must.
• Getting things you want. We want things that only other people can enable us to have. Promotions, investments, funding, volunteers, etc… When you want something from someone, it is very important to build a strong mental model about what they want, what their incentives are, and what they prioritize.

---