Substitution as the Reverse Chain Rule

Substitution is not permission to rename part of an integrand. It is the reverse of a derivative rule you already know. The Chain Rule says that differentiating a composite function requires the derivative of its inside function. Substitution reverses that structure: when an integrand contains a function of an inner expression and the differential change of that inner expression, it can be rewritten as a simpler integral. The central question is therefore not “What can I call $u$?” but “Which change of quantity makes the whole differential expression match a known antiderivative?”

Begin with the integral

$$\int 2x\left(x^2+1\right)^5\,dx.$$

There are two nested layers. The expression $x^2+1$ is the inside layer, and it is raised to the fifth power by the outside layer. Its derivative is $2x$, which is already present as the remaining factor. That pairing is the recognition cue. The integral is not difficult because the fifth power is large; it becomes simple because the integrand carries exactly the change needed to replace $x^2+1$ by one new variable.

Read the derivative structure first

The first diagram separates the two layers of the example. The blue path shows the input $x$ becoming $x^2+1$; the purple path shows that result becoming its fifth power. The orange annotation is the crucial connection: a small change $dx$ in the original input produces the change $2x\,dx$ in the inside quantity. Read that annotation as a whole. A derivative without its differential is not yet the replacement $du$.

Write the substitution and its differential together:

$$u=x^2+1, \qquad du=2x\,dx.$$

This is an equality of differential expressions, not a decorative side note. It accounts for every factor of the original integrand: $u^5$ replaces $\left(x^2+1\right)^5$, and $du$ replaces $2x\,dx$. After those two replacements, no $x$ should remain. That last sentence is a powerful error check. If a purported $u$-integral still contains $x$, then the change of variable has not been completed.

The rewritten integral is now familiar:

$$\int 2x\left(x^2+1\right)^5\,dx =\int u^5\,du =\frac{u^6}{6}+C.$$

Substitute back because the original question was written in terms of $x$:

$$\frac{\left(x^2+1\right)^6}{6}+C.$$

The constant $C$ remains because this is an indefinite integral. It represents the whole family of functions with the same derivative. Substitution did not create or remove the constant; it only rewrote the variable used to describe that family.

Why the differential is the bridge

The second diagram records the replacement as a conservation of structure. On the left, the inside quantity is $x^2+1$ and its small change is $2x\,dx$. On the right, the same quantity is simply named $u$, and its small change is $du$. The diagram is not claiming that $dx$ and $du$ are interchangeable by themselves. It shows the specific relationship $du=2x\,dx$, which is why the factor $2x$ matters.

Consider a near miss:

$$\int x\left(x^2+1\right)^5\,dx.$$

The same inside expression is visible, but its derivative is $2x$, not $x$. The method still works because a constant factor is manageable. Rewrite $x\,dx=\frac12(2x\,dx)=\frac12\,du$. Then

$$\int x\left(x^2+1\right)^5\,dx =\frac12\int u^5\,du =\frac{\left(x^2+1\right)^6}{12}+C.$$

The coefficient $\frac12$ is not an optional correction added at the end. It is required before the differential replacement is valid. The opposite mistake is also common: a learner chooses $u=x^2+1$ in the original example but writes $du=x\,dx$. That changes the mathematical statement and doubles the final derivative. Always differentiate the chosen $u$ explicitly before proceeding.

Verify by returning to the Chain Rule

The third diagram runs the reasoning backward. Starting with $\frac16\left(x^2+1\right)^6$, the outside derivative produces the exponent $6$, while the inside derivative contributes $2x$. The factor $\frac16$ cancels the exponent, leaving $2x\left(x^2+1\right)^5$. This is the same relationship that made substitution possible in the first place.

Differentiate the answer to verify it:

$$\frac{d}{dx}\left[\frac{\left(x^2+1\right)^6}{6}\right] =\frac16\cdot6\left(x^2+1\right)^5\cdot2x =2x\left(x^2+1\right)^5.$$

The derivative exactly reproduces the original integrand. This verification is especially useful when a substitution involved a constant or a power, because those are places where a lost factor is easy to hide. Notice what has been checked: not merely that the final expression “looks plausible,” but that its rate of change is the quantity the integral asked us to accumulate.

When substitution is and is not the first move

Substitution is appropriate when composition and a matching inside derivative are visible, perhaps after a small algebraic rearrangement. It is not a universal response to parentheses. For example,

$$\int x\ln(x)\,dx$$

contains a product, but neither factor is naturally the derivative of the other factor’s inside expression. Calling $u=\ln(x)$ would give $du=\frac1x\,dx$, which is not present. Calling $u=x$ simply renames the variable and does not simplify the integral. The product-rule structure instead points toward integration by parts, a later technique.

Likewise, a definite integral requires an additional decision about its bounds. You may substitute, integrate in $u$, then return to $x$ before evaluating. Or you may transform both endpoints into $u$-values and stay entirely in the new variable. Both workflows are correct only when every symbol is consistent. The next lesson makes that consistency visible rather than treating changed bounds as an afterthought.

Check your reasoning

Evaluate the following integral by first identifying the inside quantity and the factor that becomes its differential:

$$\int 3x^2\cos\left(x^3-4\right)\,dx.$$

The appropriate choice is $u=x^3-4$, because $du=3x^2\,dx$. The integral becomes $\int\cos(u)\,du$, so an antiderivative is

$$\sin\left(x^3-4\right)+C.$$

Confirm the result by differentiating it. The Chain Rule produces $\cos(x^3-4)\cdot3x^2$, which is the original integrand. If your work produced a different coefficient, return to the line where you wrote $du$; that is where the structural connection is made or lost.

Carry the idea forward

Substitution succeeds because it preserves a derivative relationship while using a simpler name for the inside quantity. Your next question is not whether the method still works for definite integrals; it does. The question is how the interval must be described after the variable has changed. The next lesson follows the same bridge from $x$ to $u$, this time with endpoints that must travel with it.