Learning curves

Today, yet again, I heard the phrase “steep learning curve.” This has always bothered me. Every time I hear this (quite overused) term, I find myself momentarily confused about what is meant. The reason for my confusion is that my notion of a steep learning curve–from studies of learning in animals and in human subjects–implies something different about the inherent difficulty of the task than what most users of the term seem to mean.

Most people who use the term “steep learning curve” are talking about something that is hard. To take just one example–one of the first I happened to encounter in a Nexis search of the term–this WaPo story is about an actress, Lauren Conrad, who is daunted by having to “learn a lot of facts and big words” for a guest-starring role on “The Family Guy.” The headline says she had a steep learning curve, and implied she found the task hard. To see whether this usage is typical, I spot-checked the Nexis results going back a few years. Every single article I opened equated steep learning curves with hard tasks.

By contrast, I tend to think of a steep learning curve as something that occurs when the task is easy. This is because a learning curve typically charts an animal’s (or human’s) response as a function of time or experience. More specifically, the learning curve shows the number or percentage of correct responses as a function of the number of attempts. Or it might show the error rate as a function of experience, in which case the learning curve would slope downward. Either way, the steeper the curve, the quicker the learner improves its performance. For example, here are graphs from a study on sequential learning by rats in which animals having to learn to press individual bars presented in a circular array in a particular sequential order.


Some versions of the task are easier for rats than others, and result in steeper learning curves.

I’ve often wondered how the jargon got co-opted. For a while my favorite hypothesis assumed three stages to the co-opting of the term. Stage 1: people were thinking something like “Wow, Lauren Conrad had a tough hill to climb in learning all those facts and big words for ‘The Family Guy’.” Stage 2: Learning curves are kind of like hills. Stage 3: Steeper hills are harder to climb (or descend safely?), and so, therefore, are steeper learning curves.

Studying instances of the term “steep learning curve” in my Lexis search, I realized another dimension of the usage: when people talk about someone having to climb a steep learning curve, they almost always mean that the task is not only hard, but it has to be done quickly.

So, there is something going on about motivation and not just the inherent difficulty of the task.

Here’s an example of different (ascending) learning curves measured when the inherent difficulty is presumably identical (a rat has to learn which bar needs to be pressed to get the reward), but the motivations are different.

Rats learn quicker when more motivated

Rats learn quicker when more motivated

The motivator in this case is a drug that is delivered when the correct bar is pressed. The higher the dose of the drug, the steeper the learning curve. See the original source of this figure in Addiction Science.

So I think I have a new insight about the co-opting of the term “steep learning curve.” When someone needed to talk about how many facts and big words Lauren Conrad had to learn, what was of interest was not really how inherently difficult the task was to achieve, but how urgently Lauren needed to achieve it. This kind of usage, then, is obviously connected with the motivations of the person doing the learning, whether it is money, fame, or drugs.

Understanding the usage in this way doesn’t dispel my discomfort when I hear the term. The problem is that the relationships among learning and motivation and performance are really complicated. A quick improvement in performance could be a result of the task being easy or the learning being very motivated. Saying the learning curve is steep doesn’t indicate which is which. But I have no intention of trying to change this incredibly common usage pattern. This would be like trying to get people to pronounce “forte” or “finance” or “dissection” the way I was taught. Some hills are too steep to climb.

As for the phrase “going forward”……


Update: Wikipedia has a decent review of the science and language of learning curves: http://en.wikipedia.org/wiki/Learning_curve

Update 10/6:  corrected some typos


About ethologist

Professor in the Department of Integrative Biology at Michigan State University
This entry was posted in Ethology, Jargon. Bookmark the permalink.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s