Pattern Recognition of the Human Brain
Like every MBA student, I find myself in endless “case cracking” workshops — preparing for management consulting case study interviews.
DEDUCTIVE LOGIC
Case studies, and generally management consulting practices, require deductive thinking: First, define assumptions, constraints, and the problem with extreme prejudice — based on the client’s interests and needs. Then, explode the problem into logical branches in an issue or logic “tree.” Follow branches linearly until you cannot go further. Apply analysis, expand the tree, eliminate infeasible branches, establish a hypothesis based on probability (return vs risk) until a single optimal branch remains: this is your client recommendation.
You can make it through this process using logical tools — no prior knowledge of the client is required. Careful, the tree is a tool to comprehensively explore problem factors. If the problem statement did not incorporate accurate assumptions and constraints, recommendations are not even solutions, merely logically derived thoughts.
If this is how you naturally think, great! You may join the 1% of applicants that receive offers and start your sexy new management consultant job at $150k/year+ with annual 20% raises in New York City.
But what if you don’t think this way? Well, better get case cracking with your cohort. Maybe 100 practice cases will make you good enough to get the job.
INDUCTIVE LOGIC
Let’s assume that you are an intelligent, strategic, competent professional problem solver, but not a deductive thinker. By what methods do you solve problems? Chances are, you are an inductive thinker.
Inductive logic leads to a hypothesis through evidence, rather than a principal leading to deduction. A simple example: if you see three white swans, you might induce a fourth swan will also be white. However, if the world-wide authority on bird colors wrote all swans are white, then you will deduce the fourth swan is white (without needing to see the other three).
Induction is an enormously effective method of solving ambiguous problems quickly. When you know there is a problem but are not ready to define it, inductive thinking organizes potential hypotheses through pattern recognition.
I can’t say I’m an expert on inductive thinking. Inductive and abductive logic are the basis of design thinking, which I have never formally studied. I only have myself as the case study (pun intended), which is to say, I do this naturally in most settings; I didn’t learn it.
I solve problems by absorbing disparate, detailed and high volume data, associating the new data to new or other data, and finally applying the associations in a novel conclusion. A more concrete comparison: I’m a pattern recognition machine.
Artificial intelligence — an actual pattern recognition machine — generally involves teaching a computer a pattern and its meaning, for example, teaching a machine to identify suspicious behavior to prevent or catch shoplifters. You feed the machine a ridiculous quantity of handwriting examples and the machine will learn to read handwriting (or maybe write it!):
INDUCTIVE LOGIC EXAMPLE
I’ll walk you through an example problem from my time at SpaceX.
A supplier calls to say a part you need today (really, a few weeks ago) is out of tolerance. I had a clear and vague mandate — do anything and everything necessary to launch rockets.
A deductive thinker might quick around a few problem statements, question why the part is out of tolerance, do some analysis, conduct some interviews, prioritize a few specific options, select the most optimal (fastest, lowest cost, highest likelihood of success), and implement. A deductive thinker might rely on a toolkit or set of potential procedures. Likely, the solution set would include repair, replace, or do without, and a few ways to do each one.
Here’s the issue. When you needed something weeks ago, it’s difficult to have the creativity necessary to try things that don’t have solid logic reason backing them. My goal was to creatively break every law of physics and every rule or protocol (within legal/ethical bounds) to make a physical part appear on time to integration.
This means that I don’t actually need that exact part to be the one integrated. I need something. But I really don’t understand what. The part might need to be redesigned. I might use a different part. I might need multiple solutions in case some fail. If you start with a problem statement of “how do I get this part to integration on time” you unnecessarily limit yourself. Or, Elon fires you before you craft a masterful problem statement encompassing every potential solution set ever.
The real solution set includes EVERYTHING and NOW.
Practically, how would I go about determining “everything?”
Figure 1 shows the sum quantity of data I know. Each dot represents an observable event, principle, knowledge, etc., across the four dimensions, including time.
- The details of the defect (if any)
- Known supplier performance
- Alternative suppliers, including internal
- Alternative production methods
- Alternative parts
- Alternative materials
- Production schedule and feasibility (which it isn’t lol), both upstream and downstream
- True integration timing (aka when my customer needs the part)
- Engineering organizational structure and who are most helpful
- Previous crises and crises responses
- The necessity of the part (aka does rocket science say we need the part for the rocket to go up)
- Rework feasibility
- Production steps
- SpaceXers who have skills to fabricate this part
- Machine types needed to produce, including alternates
- How long it takes to rewrite process planning
- Transit length between supplier(s) and SpaceX
- Spare raw material
- Inventory balance
- What Elon had for lunch
- How politely Elon’s kids ask for fro-yo at the LunchPad. So cute.
- Etc. etc. etc.
I don’t spend a second defining the problem. It seems pretty clear — I’m supposed to launch rockets, and I’m being prevented. I am thoroughly uninterested in understanding the root cause of the defect and any permanent consideration— I’m wholly focused on immediate solutions.
Again, each data I know is represented by a dot. Several patterns are immediately obvious, essentially clusters that represent potential solutions. I’ve drawn a visual representation of patterns in Figure 2.
- Previous supplier performance is bad, so my hypothesis is that the supplier will be unable to fix the part in time; therefore, I will immediately insource the part in our machine shop and dual source at multiple suppliers — expedite all.
- I have a good relationship with the responsible engineer. I know we have crises all the time, but this guy is good at understanding the risk of the defect on the mission. I’ll see if he’ll open a risk ticket (aka fly the mission without the part or with the defect) or if he’ll suggest an alternate part, rework plan, etc.
- So on and so forth…
About 8 ideas come to mind of how to extricate us from the situation as soon as possible. Speed and creativity is of the essence, so I sought and found sufficient patterns within the data itself. As a good SpaceXer, I then put all in motion. Especially since I know the next 3 parts from the supplier are also late — if we can get maybe 4 of the solutions to work, we’ll be caught up to schedule with some buffer! Now, onto the next crisis…
BECOMING AN INDUCTIVE THINKER
At SpaceX in my example, there were many potential solutions. But at the beginning, I didn’t know how to describe an optimal problem statement and I wasn’t sure how to verify which solutions will actually work in time. This means deductive logic might help clarify thoughts or communicate potential solutions clearly (more on this later), but it cannot solve the “situation” because it is only useful for a single problem statement.
The process for improving inductive thinking is lifelong. I personally read prodigiously to gather as much data as possible. Then, I create temporary or situational structures by identifying patterns. Finally, I add repetition to train and retrain the brain in pattern recognition.
Inductive thinking is not disorganized, as it deals with forming organized patterns. Therefore, it should not be confused with disorganized thinking, internal vs. external processing, etc. Structure is crucial to inductive thinking, as it solidifies patterns for the purpose of decision making, much as a database key links disparate databases.
An example of structure: Christians value forgiveness — as a Christian, if I encounter a situation given the choice between forgiving and not forgiving, I chose to forgive; in this way, structure could be called a principle or worldview. Deduction would say, “Jesus said to forgive your neighbor,” and induction would say, “Jesus forgave everyone who wronged Him.” They are not quite the same statements, but often result in the same decision (to forgive).
In inductive thinking, the structure isn’t initially clearly defined (Elon wants SpaceX to launch rockets at any cost… but which costs are acceptable? What if we need to land rockets too? Or create revenue streams outside of rockets to fund rocket launches?), so structure must be created from patterns. Once the structure is established, hypotheses are relatively easy to create, lending to speed in decision making.
THE CHALLENGE OF INDUCTIVE LOGIC
Inductive thinking must be obsessively cultivated through learning to improve the the ability to properly assess patterns, just as a machine is fed new data continually. This means… you should learn to think deductively. ;)
As inductive thinking produces hypotheses derived from evidence, communicating an argument is more challenging than in deductive logic.
The reality is that deductive thinkers rule the world. Government officials, business executives, etc., all have roles where decision must be made with precision for a specific purpose.
The issue with inductive logic is that it relies on rational hypotheses rather than logical deduction based on accepted theory or fact. Inductive logic is great for on-the-fly problem solving with like-minded individuals or smaller problems where the problem statement is assumed. Outside of the proper context, it is unlikely that the pattern clusters a person identifies internally will be readily understood. Or, for problems that require “getting it right” with extensive research and selling of a solution, inductive reasoning isn’t going to help sufficient consideration of all related factors. If observable data is incomplete or inaccurate, the likelihood is high that you will waste time chasing a useless solution.
Figure 3 shows alternative patterns to Figure 2. Did I miss crucial solutions if I saw only 2 patterns instead of 3? Are these patterns more likely to produce better hypotheses?
Inductive thinking is more readily understood ambiguous situations, because it is expected. However, inductive thinking relies on organizing data into trends to formulate a hypothesis. This is a highly subjective process, and hard to communicate, especially depending on communication preferences.
A highly developed inductive thinker will communicate hypotheses with precision, including assumptions and constraints. A less developed inductive thinker might spit out every potential problem statement and solution without particular order and organization, worst case scenario, not even realizing he has become a walking, talking logical contradiction. It’s all in his head.
Very few are exploring potential problem statements, with iterative (and expensive) trials on the path of determining a problem and solution. That’s why inductive thinkers are most likely found in early stage ventures or creative roles. That’s also why inductive thinkers, like me, face a structural disadvantage in interviewing for higher level roles… We seemingly don’t make sense.
Try to convince an investor to give you millions of dollars that seemingly random details add up to a pattern — I doubt you’ll succeed. Management consulting firms test deductive logic ability in interviews for precisely this reason — they sell ideas, not solutions. Clients must trust the derivation of ideas or else even the best ideas will be ignored.
My advice is this:
- Know which problems need deductive logic, and which need inductive logic
- Learn multiple communication styles and logical methods through constant practice
- Even after using inductive logic to identify a solution, sell the solution using deductive logic. That is, go backward, define the problem statement, clarify assumptions, branch out trees (because you need to be ready when people ask if you considered X and Y, and they will). Let the narrative follow the logic
- Tell people when you are using inductive logic. Most people are natural deductive thinkers. If you start considering ideas trying to sort out patterns without yet committing to a hypothesis, and don’t tell anyone, most people will think you are committing to what you say.
THE CHALLENGE WITH DEDUCTIVE LOGIC
Here’s an application of logical methods: if inductive thinking is best for solving ambiguous problems, then deductive thinking is the best for solving structured problems. Therefore, inductive thinking is not useful for solving structured problems and deductive thinking is not useful for solving ambiguous problems.
Okay, that wasn’t good logic because the inverse of a logical statement isn’t necessarily true. But, my hypothesis, derived inductively, is that this logical statement is true. Deductive thinkers impose constraints on themselves that prevent creativity.
An example: a case study in marketing class asked into which of 3 cities Cirque du Soleil should expand in 2007. Data on the 3 cities was provided. Of the 8 groups in the class, 2 would be allowed to present for extra credit. Interested groups would send a pitch via email to the professor. The first group chosen to present would be the first group to email the professor. The second group chosen would have a different recommendation than the first group. I figured my group would not win in speed, so I recommended that we sell Cirque to Disney. We were selected to present with much enthusiasm by the professor. I had to break the rules of the defined problem to succeed.
High growth startups attract inductive thinkers because creativity outside of established, logical boundaries is necessary to create products to in new markets or to disrupt rich, established competitors. SpaceX cared more for recruiting thinkers who could recognize patterns in any situation than accomplished aerospace professionals because the SpaceX model is fundamentally different than the established aerospace model. That’s not to say SpaceX didn’t hire from management consulting firms like McKinsey: certain problems, such as process implementation at economies of scale, are best done removed from chaos, with careful consideration of all factors to solve the right problem.
CONCLUSION
Now that I’ve discussed definitions-by-example of inductive and deductive logic, potential consequences, let’s tie this back to case studies. I’m in France getting a top tier MBA to learn deductive methods so I can become a balanced problem solver.
Deductive and inductive thinkers will bring different strengths to case studies. As you practice case crackers, emphasize the strengths but flip the case upside-down every once in a while.
Deductive thinkers
You will be best at establishing a clear problem to solve. Your failing will be applying a structure to a problem when it doesn’t fit and therefore constraining yourself from optimum solutions. In other words, you’re boring.
Pink, fluffy ponies. Break the rules. Read up on startups, disruption, blue ocean strategy. Make some logical leaps. Feed yourself an endless repetition of creativity until you can recognize rational and meaningful extrapolations outside the traditional problem constraints.
Inductive thinkers
You need to learn how to organize your thoughts. This is difficult because establishing the framework is exhausting and you’ll probably forget half your ideas in the process. Again, feed yourself good patterns by observing great examples of case cracking. Ask lots of questions about the motivations for structuring a problem one way or another, as it won’t be abundantly clear why the framework matters until someone super smart can convince you of the “why” behind the “what” (after all, you solved the problem a long time ago and are pretty bored with having to convince the interviewer your solution). Eventually, you’ll start to become comfortable with frameworks. When you see your salary jump through endless promotions, you’ll appreciate your effort.
Merci beaucoup et bonne journée!
Did I miss any crucial patterns? Let me know below.
