If you're seeing this message, it means we're having trouble loading external resources on our website.

If you're behind a web filter, please make sure that the domains *.kastatic.org and *.kasandbox.org are unblocked.

Main content

Applied rate of change: forgetfulness

Let's face it—sometimes we forget things! Ever wonder about the dynamics underpinning this phenomenon? Let's explore this by tracking the number of vocabulary words remembered over time. Together, we'll use a mathematical model to calculate the rate of change in our memory, revealing the fascinating dynamics of knowledge retention and loss. Created by Sal Khan.

Want to join the conversation?

  • starky ultimate style avatar for user Brandon
    Where did the modelling function come from? How is it known that this particular function models the forgetting of words?

    This has popped up into my mind before... I always wondered where the parabolic path functions came from for physics, and I found that they are the results of diff eqs from manipulating the definitions of acceleration and velocity.

    Is there some definition and diff eq that will yield this function as well?

    In general, is there a class that teaches you how to model things? I can do it easily with geometric problems, however, it is not so easy for me with things like this, but once I figured out the diff eqs thing and the definitions, it was cake. I really would like to know how to make models of anything I see.
    (19 votes)
    Default Khan Academy avatar avatar for user
    • leafers sapling style avatar for user Peter Collingridge
      I'm almost certain that the modelling function in this question was made up purely for this question.

      There are two main ways that science comes up with models/functions. The first empirical, which means you do an experiment. You get a large sample of people, teach them some words and then test their ability to recall then after certain time periods. Then you could plot the number of words over time on a graph and try to fit a function. For linear functions you can use linear regression to find the equation of the line that best fits the data. For non-linear functions it's more tricky, but there are techniques. Even then you often have to make a judgement call about whether you think the function should be polynomial (e.g. y = x^4) or exponential (e.g. y = 2^x) or something else. That's when you might want to think of a theory to explain the function.

      The second way would be theoretically. Here you basically start with a function that you think is logical. Or you start with a simple model of the world and see what functions come out of that. That's tricky in this example, but more common in physics. Maybe you could imagine the brain is a bucket containing information and every day it randomly loses a proportion of the information in it. This would give an exponential decay type function (and probably isn't very accurate). Then you should go to the experiment approach and see if your model accurately predicts reality.

      As for where the parabolic trajectories come from in physics: it's a result of gravity imposing an effectively constant acceleration on objects near the surface of Earth. A constant acceleration means that the downward velocity of an object increases linearly. And this means the position of the object above the surface decreases with the square of time. The velocity in the horizontal direction is assumed to be constant (i.e. there's no friction), so as the x-coordinate of an object changes linearly with time, the y-coordinate changes with the square of time.

      I hope that makes some sense. don't think there's anything on Khan Academy that teaches this, but you could try: ://www.coursera.org/course/modelthinking
      (64 votes)
  • ohnoes default style avatar for user Cyan Wind
    The rate of change of the number of known words per day... What does the word per day means? w'(2) = -12.8, w'(3) = -11.2 and w'(4) = -9.6. I know that the number of words (w(t)) will keep decreasing but why per day?
    (8 votes)
    Default Khan Academy avatar avatar for user
    • purple pi purple style avatar for user doctorfoxphd
      We are talking about a unit rate that changes every day according to the curve we were given. Unit rates often include time as the independent variable, such as miles per hour, meters per second, words per day. In the case of this forgetful guy, the curve is a parabola that says he forgets fastest at first and slower later on. Here are some examples:
      Day Loss rate
      1 -16
      2 -12.8
      4 -9.4
      5 -8
      9 -1.6
      10 0 (he has already forgotten all of them by then)
      So, by the 9th day he is forgetting words at only 1.6 words per day. Too bad that wasn't his forgetting rate at the beginning--he would ace his test.
      (9 votes)
  • duskpin tree style avatar for user Zor
    I took the equation 80 (1-0.1t)^2 and expanded it and then took the derivative of that and got a totally different number. what's up with that?
    (3 votes)
    Default Khan Academy avatar avatar for user
  • blobby green style avatar for user Peter Lock
    I am trying to solve the second derivative of
    X^2+y^2= 6xy
    Can you help
    (1 vote)
    Default Khan Academy avatar avatar for user
    • duskpin sapling style avatar for user Vu
      x² + y² = 6xy
      2x + (2y)y' = 6*y + 6x*1*y'
      (2y)y' = 6y + (6x)y' -2x
      (2y)y' - (6x)y' = 6y - 2x
      y'(2y - 6x) = 6y - 2x
      y' = (6y - 2x) / (2y - 6x)
      y' = [2(3y - x)] / [2(y - 3x)]
      y' = (3y - x) / (y - 3x)

      y" = [(y - 3x)*(3*y' - 1) - (3y - x)*(1*y' - 3)] / (y - 3x)²

      That's our second derivative, but we have y' in it. We do know what y' is so we can substitute y' = (3y - x) / (y - 3x) in for y'. Then just simplify it. Note I do all this on the computer and with all the nesting, it's very confusing, so there is a chance that I made some arithmetic error. But the point is substitute in the first derivative for y' and simplify.

      y" = [(y - 3x)*(3*y' - 1) - (3y - x)*(1*y' - 3)] / (y - 3x)²
      y" = { (y - 3x) * {3 [(3y -x) / (y - 3x)] - 1} - (3y - x) * { [(3y - x) / (y - 3x)] - 3} } / (y - 3x)²
      y" = { [(9y - 3x) - (y - 3x)] - [((9y² - 6xy + x²) / (y - 3x)) - (9y - 3x)] } / (y - 3x)²
      y" = { (9y - 3x - y + 3x) - [((9y² - 6xy + x²) - (9y² - 30xy + 9x²)) / (y - 3x)] } / (y - 3x)²
      y" = {8y - [(9y² - 6xy + x² - 9y² + 30xy - 9x²) / (y - 3x)] } / (y - 3x)²
      y" = { [8y (y - 3x) / (y -3x)] - [(24xy -8x²) / (y -3x)] } / (y - 3x)²
      y" = { [(8y² - 24xy) - (24xy - 8x²)] / (y -3x) }/ (y - 3x)²
      y" = [(8y² - 24xy - 24xy + 8x²) / (y -3x)] / (y - 3x)²
      y" = (8y² - 48xy + 8x²) / (y - 3x)³
      y" = [8 (y² - 6xy + x²)] / (y - 3x)³
      y" = 8(x² + y² - 6xy) / (y - 3x)³

      Note that our original equation is x² + y² = 6xy, we can substitute in 6xy for x² + y².

      y" = 8(6xy - 6xy) / (y -3x)³
      y" = 8(0) / (y -3x)³
      y" = 0
      (6 votes)
  • primosaur seed style avatar for user Luky
    what is the meaning of the rate of change number itself? it can be probably used to draw a secant on the graph and to decide if the value on the graph is decreasing or increasing but besides that?
    (3 votes)
    Default Khan Academy avatar avatar for user
  • mr pink red style avatar for user svk2799
    what is the formula of rate of change probllems
    (1 vote)
    Default Khan Academy avatar avatar for user
  • mr pink red style avatar for user andres abraham
    why is the constant not subjected to the derivative, shouldn't the derivative of a constant always be zero? At
    (1 vote)
    Default Khan Academy avatar avatar for user
  • blobby green style avatar for user MOHALE
    can we use the gradient formula to get the rate instead of using the derivative.
    (1 vote)
    Default Khan Academy avatar avatar for user
  • leafers ultimate style avatar for user Akash Thakur
    the term (-0.1t) ensures the negative slope ;but with t(+ve number) multiplied with it
    we are making slope less negative?? is this a good model ?if anyone knows any other model of this problem please tell me ,
    (1 vote)
    Default Khan Academy avatar avatar for user
  • piceratops ultimate style avatar for user Sree
    Sal's question is a classic case of our brain skipping the the second the. (reread this comment and the question)
    (1 vote)
    Default Khan Academy avatar avatar for user

Video transcript

I studied for an English test today and learned 80 vocabulary words. In 10 days, I will have forgotten every word. The number of words that I remember t days after studying is modeled by-- so W of t, so this is the number of words I have in my head as a function of time is going to be equal to 80 times 1 minus 0.1t squared for t is between 0 and 10, including the two boundaries. That's why we have brackets right over here. What is the rate of change of the number of known words per day two days after studying for the test? And I encourage you to pause this video and try it on your own. So the key here is we come up with this equation for modeling how many words have retained in my brain every day after I first memorized them, after I got the 80 of them into my head? And that's this expression here. And they want to know the rate of change two days after studying. Well, the rate of change, I can take the derivative of this with respect to time. So let's do that. So let's take the derivative. The derivative of the number of words I know with respect to time is going to be equal to-- well, we have this 80 out front. That's just a constant. And now I can apply the chain rule right over here. So the derivative of 1 minus 0.1t, the whole thing squared with respect to 1 minus 0.1t is going to be-- so I'm essentially taking the derivative of this whole pink thing, this whole expression squared with respect to the expression. So that's going to be 2 times 1 minus 0.1t. And now I can find the derivative of this inner expression with respect to t. So the derivative of this inner expression with respect to t is just going to be 0 minus 0.1. So it's just going to be negative 0.1. And of course, we can simplify this a little bit. This is going to be equal to-- if we take 80 times 2 is 160 times negative 0.1, that's going to be negative 16. 160 times 0.1 is 16, so negative 16 times 1 minus 0.1t. And if we want, we could distribute the 16, or we could just leave it like this. But we're ready now to answer our question. We could write this as the rate of change of the number of words we know with respect to time. Or we could use the alternate notation. We could say this is W prime of t. Either way, it's going to be equal to this thing. Let me do that same color. It's equal to negative 16 times 1 minus 0.1t. So what's this going to be? What is the rate of change of the number of words known per day two days after studying for the test? Well, we just have to evaluate this when t is equal to 2. So W prime of 2 is going to be equal to negative 16 times 1 minus 0.1 times 2 close parentheses. And that's going to be equal to-- well, let's see, what is this? This is 1 minus essentially 0.2. This is going to be 0.8. So this is going to be equal to negative 16 times-- is that right? 1 minus 0.2 is-- yep, it's going to be times 0.8. And what is that going to be? If I were to multiply 16 times 8, it would be 128. It's 2 times 8 times 8 so 2 times 64, 128. So this is going to be negative 12.8. And just to really hit the point home of what we're doing, the rate of change is negative 12.8 words per day. So if you believe this model for how many words we know on a given day, this is saying on day two, right at the day two point, right after exactly two days after studying for the test, right at that moment, I am essentially losing 12.8 words per day. The number of words I know is decreasing by 12.8 words per day.