You find where a function is decreasing by scanning the graph left to right and marking every stretch where the curve falls as x increases. On a typical precalculus or AP Calculus graph, that means reading the x-values off the horizontal axis, not the y-values off the vertical one. The single most common error is writing the interval in y-coordinates or using open brackets where the textbook expects closed ones. Both mistakes cost points on free-response questions even when the shape was read correctly.

Read the Decreasing Interval as an X-Range, Never a Y-Range

A decreasing interval is a set of x-values, written in interval notation with the lower x on the left. If a curve falls from the point (1, 7) to the point (4, 2), the interval is (1, 4) or [1, 4] depending on your course's convention β€” the 7 and the 2 never appear in the answer. Students who write "from 7 to 2" have described the vertical drop, not the interval.

Think of walking the graph like a hiker heading east. Your horizontal position is x; your altitude is f(x). You are descending whenever your altitude drops as you keep moving east. The question "where is the function decreasing" asks for the stretches of trail where you lose elevation, reported in miles east, not in feet of altitude lost.

This distinction matters because the same decreasing interval can belong to wildly different vertical drops. A line from (0, 100) to (1, 0) and a line from (0, 1) to (1, 0) both decrease on (0, 1). The interval describes the input, and the input alone.

One quick self-check: if your answer contains a number larger than the rightmost tick mark on the x-axis, you probably grabbed a y-value by mistake. Re-read the horizontal axis labels before writing anything down.

Find the Turning Points First, Then Test the Stretches Between Them

Decreasing behavior only changes at specific landmarks: peaks (local maxima), valleys (local minima), and breaks such as vertical asymptotes or jump discontinuities. Locate those first, and the graph splits into clean candidate stretches that are easy to test one at a time. This turns a whole-curve question into three or four small yes/no questions.

For a smooth polynomial like a cubic with a peak at x = βˆ’2 and a valley at x = 1, the candidate stretches are everything left of βˆ’2, the stretch between βˆ’2 and 1, and everything right of 1. Test one point inside each stretch by comparing the height at that point to the height slightly to its right. Falling height means decreasing.

Rational functions add a wrinkle: a vertical asymptote at x = 3 splits the domain, so "decreasing on (βˆ’βˆž, 3)" and "decreasing on (3, ∞)" are two separate statements. Some textbooks let you union them with a comma; others refuse because the function isn't defined at 3. Ask your instructor which convention applies before the exam, because both appear in print.

Piecewise graphs can decrease, then jump upward, then decrease again. The jump itself is neither increasing nor decreasing β€” it's a discontinuity β€” so the intervals stop on either side of the jump and never span it.

Walk Through Four Steps to Lock In Each Interval

Here is the working procedure I'd hand a student before a quiz. It takes about 90 seconds per graph once you've practiced it twice.

1: Trace the curve left to right with your finger or a pencil tip. Say "down" out loud each time your fingertip drops as it moves right. The drops are your decreasing stretches. 2: Mark the x-coordinates where the direction changes. A peak (high point) ends a decreasing stretch going left-to-right; a valley (low point) begins one. Write those x-values on scratch paper. 3: Test one sample x inside each candidate stretch. Pick a value like x = 0 or x = 2, read f at that x and at x + 0.1, and confirm the second reading is lower. If it is, the whole stretch is decreasing. 4: Write the interval in x-notation and check the bracket style. Open brackets ( ) mean the endpoint is excluded; closed brackets [ ] include it. Many courses accept either at turning points, but AP Calculus free-response rubrics have historically accepted both open and closed endpoints for monotonicity intervals β€” confirm with your teacher's rubric.

Failure handling: if your finger moves down but the interval you wrote doesn't match, you almost certainly swapped x and y. Re-read step 1 and start again. If the graph has a flat plateau, that stretch is neither increasing nor decreasing, and it should be excluded from your answer or noted separately.

If the graph is hand-drawn and ambiguous near a turning point, plot two or three extra points in a table of values. A quick numeric check settles what the eye can't.

Slope Sign Beats the Visual Drop Every Time

A function is decreasing exactly where its derivative is negative β€” that is, where the tangent line slopes downward. On a graph without an equation, "slope is negative" translates to "the curve falls as you move right." This is the formal definition your calculus course will use, and it matches the visual read every time for differentiable functions.

The definition in most calculus texts, including standard AP-aligned ones, is: f is decreasing on an interval if for every pair a < b in that interval, f(a) > f(b). That's a comparison of outputs across the whole interval, not just at a single point β€” which is why a function can have a negative slope at one point yet not be decreasing on any interval around it.

Steepness tells you how fast the drop happens, not whether it's a drop at all. A very shallow downward slope is still decreasing. A steep upward slope is still increasing. Don't let the visual drama of a steep section fool you into calling it decreasing just because it looks energetic.

Where the slope is exactly zero at an isolated point β€” like the flat moment at the top of a hill β€” that single point belongs to neither category. It's a turning point, and it's the boundary between an increasing stretch and a decreasing one.

Open Versus Closed Brackets at Endpoints, and Why Courses Disagree

Textbook conventions split here, and the split is real. Some precalculus texts, including several widely used Larson and Stewart editions, define monotonicity on open intervals and report decreasing intervals with parentheses. Others accept closed brackets because the definition using f(a) > f(b) technically holds at the endpoints of a maximal interval. The result is that identical correct work can be marked right in one class and half-right in another.

Convention Notation Typical source
Open intervals only (a, b) Many precalculus texts; some online homework systems
Closed at turning points [a, b] Some calculus texts; certain AP rubrics
Either accepted (a, b) or [a, b] Common in AP Calculus free-response scoring

The practical rule: write what your teacher writes on the board, and if you're taking a standardized exam, follow the published scoring guidelines for that exam year. The College Board has accepted both open and closed endpoints on monotonicity questions in past AP Calculus rubrics, so the safest play is to match your instructor's classroom convention unless the rubric says otherwise.

One more bracket subtlety: ∞ and βˆ’βˆž always take parentheses. You never write [0, ∞) with a closed bracket on infinity, because infinity isn't a number you can include.

If you're using an online homework platform like WebAssign or MyMathLab, check a worked example in the platform's own help pages before your first submission. The system's grader enforces one convention, and it may not match your lecture notes.

Discontinuities, Asymptotes, and Jumps Break the Interval

A decreasing interval must be a connected stretch of the domain. If the graph jumps, has a hole, or shoots off toward a vertical asymptote, the decreasing behavior stops there and a new interval begins on the other side. Writing a single interval across a break is one of the most common errors on rational-function problems.

Take f(x) = 1/x. On (βˆ’βˆž, 0) the curve falls from near zero toward βˆ’βˆž, and on (0, ∞) it falls from +∞ toward zero. Many textbooks describe this as "decreasing on (βˆ’βˆž, 0) and on (0, ∞)" but refuse to say "decreasing on (βˆ’βˆž, 0) βˆͺ (0, ∞)." The reason: the definition requires f(a) > f(b) for every a < b in the interval, and picking a = βˆ’1 and b = 1 gives f(βˆ’1) = βˆ’1 and f(1) = 1, which violates the condition.

Piecewise functions behave the same way. A jump upward followed by a downward slope is two separate stories. Note the jump point as excluded from either interval.

Vertical asymptotes at x = c mean the interval endpoints approach c but never reach it. Write (a, c) and (c, b), never (a, b) with c buried inside. If the asymptote sits at the edge of the domain, the interval simply ends there.

When in doubt, sketch the domain on a number line first. The breaks in the domain dictate where your interval list must split.

Match the Interval to the Function Family You're Looking At

Different function families produce recognizable decreasing patterns, and knowing the pattern lets you sanity-check your interval in seconds. This is the shortcut experienced students use to catch errors before turning in the paper.

Function family Typical decreasing pattern Sanity check
Linear, negative slope Entire domain, one interval Slope is constant and negative
Quadratic opening upward Left half, from (βˆ’βˆž, vertex x) Vertex is a valley
Quadratic opening downward Never on the whole domain (only after the peak, briefly) Check the sign of the leading coefficient
Cubic with two turning points Middle stretch between peak and valley Peak sits at higher x than valley
Exponential decay Entire domain Base between 0 and 1
Sine or cosine Alternating stretches, one per half-period Period length matches the graph

Quadratic pattern is the one students misread most often. A parabola opening upward (positive leading coefficient) decreases on the left half and increases on the right; a parabola opening downward increases on the left and decreases on the right. The sign of the leading coefficient tells you which half is which without any plotting.

For trigonometric graphs, the period is the giveaway. Count the horizontal distance between two consecutive peaks; that's the period, and one full period contains exactly one decreasing stretch and one increasing stretch (for sine and cosine). Any "decreasing" interval longer than half a period is a reading error.

Exponential decay looks decreasing everywhere but never crosses zero. The interval is the whole domain, but if the graph is drawn with an asymptote at y = 0, don't mistake the asymptote for a turning point.

Worked Example: A Cubic With Two Turning Points

Let's run the full procedure on a concrete graph. Suppose the curve rises from the far left, peaks at (1, 4), falls to a valley at (3, 0), then rises again toward the right. The question asks where it's decreasing.

Step 1: trace left to right. The fingertip rises until x = 1, drops from x = 1 to x = 3, then rises after x = 3. The decreasing stretch is the middle segment.

Step 2: the peak sits at x = 1 and the valley at x = 3. Those are the boundaries.

Step 3: test x = 2. Reading the graph, f(2) β‰ˆ 2 and f(2.1) β‰ˆ 1.87. Height fell, confirming decrease on the stretch.

Step 4: report the interval as (1, 3) or [1, 3] depending on your course convention. In AP style, either is typically accepted; in a strict precalculus class, follow the instructor's call.

A quick derivative check: if the cubic is f(x) = xΒ³ βˆ’ 6xΒ² + 9x + 4, then fβ€²(x) = 3xΒ² βˆ’ 12x + 9 = 3(x βˆ’ 1)(x βˆ’ 3). The derivative is negative between 1 and 3, confirming the decreasing interval (1, 3). Setting the derivative to zero recovers the turning points at x = 1 and x = 3, which is exactly what the visual scan found.

This is why AP Calculus lets you solve monotonicity questions either visually or by derivative sign chart. Both routes give the same interval, and on the exam you can pick whichever is faster for the given problem.

Frequently Asked Questions

Does a decreasing interval include the turning points at its ends?

Most courses include the endpoints at turning points, writing [1, 3], but many precalculus texts use open parentheses (1, 3). The College Board has accepted both for monotonicity questions in past AP Calculus scoring, so check your instructor's convention before the test rather than assuming one.

What if the graph has a flat horizontal stretch?

A flat stretch is neither increasing nor decreasing β€” the function is constant there. Exclude that stretch from both interval lists, or note it separately as constant. If the constant stretch sits between a decreasing stretch and an increasing one, the decreasing interval ends at the left edge of the flat stretch, not at the middle.

Is a steeper line more decreasing than a shallower one?

No. Steepness measures how quickly the function changes, not whether it's decreasing. A line with slope βˆ’0.1 is still decreasing, just more slowly than one with slope βˆ’5. The sign of the slope determines whether the function is decreasing; the magnitude only tells you the rate.

Does the interval change if I flip the graph upside down?

Yes, every decreasing interval becomes an increasing one when you reflect the graph across the x-axis. If f decreases on (1, 3), then βˆ’f increases on (1, 3). This is useful for quick checks: any function you can relate to a known one by a vertical flip or shift inherits a predictable monotonicity pattern.

Can a function be both increasing and decreasing on the same interval?

Only if it's constant β€” and a constant function is neither, strictly speaking. If you see a stretch that seems to go up and down within the same interval, you've likely missed a turning point in the middle. Recheck the landmarks; the interval should split at every peak and valley.

Do I need calculus to determine decreasing behavior from a graph?

No. The graph-reading method (trace left to right, locate turning points, test sample points) works without any derivative. Calculus gives an algebraic shortcut β€” check the sign of fβ€²(x) β€” but on a visual graph the pure-graph method is usually faster and less error-prone.

What happens if the graph is decreasing on two disconnected stretches?

Report each stretch as its own interval, separated by "and" or a comma. For f(x) = 1/x, the standard answer is "decreasing on (βˆ’βˆž, 0) and on (0, ∞)." Avoid writing a single union across a break unless your instructor explicitly allows it, because the formal definition fails at points that straddle the discontinuity.