How to Determine Intervals Where a Function Is Decreasing From a Graph: Spot It by Slope Direction and Open Endpoints

To find where a function is decreasing from its graph, trace the curve from left to right and mark every stretch where the graph goes downhill as x increases; then write those x-values as open intervals, such as (1, 4), excluding the turning points themselves. The key condition is direction of travel, not whether the y-values are positive or negative. A curve can sit entirely below the x-axis and still be increasing. This guide walks through the trace, the interval notation, the endpoint rules your textbook or exam board expects, and the traps that cost marks.

Read Decreasing as Downhill Travel From Left to Right

A function is decreasing on an interval when, for any two x-values a and b in that interval with a < b, the function value at a is greater than the function value at b. On a graph, that condition looks like a curve descending as your eye moves rightward. You don't need the equation to see it. The visual test is direct: pick a point, move right, and check whether the new point sits lower.

Watch for the common misread. Students often scan a graph top-to-bottom and call a steep left side "decreasing." Direction matters, and the reference axis is always the x-axis moving left to right. A line falling from upper-left to lower-right is decreasing. A line rising from lower-left to upper-right is increasing. A horizontal stretch is neither.

Local behavior can flip quickly. A cubic such as f(x) = x³ − 3x increases, turns, decreases, and turns again. Each downhill stretch is its own interval. You report all of them separately rather than merging them into one span, because the upward piece in the middle breaks the pattern.

Use the tangent-slope picture as a second check. Where the graph descends, tangent lines have negative slope. That connects the visual reading to the derivative test you'll meet in calculus, where f′(x) < 0 on the same intervals. The graph and the algebra should agree.

Trace the Curve With a Left-to-Right Pencil Sweep

This is the procedure most teachers expect, and it takes under a minute on a clean graph.

1: Draw or imagine a vertical line at the far left of the graph and slide it rightward across the entire visible domain. 2: At each position, note whether the curve the line crosses is moving down (decreasing), up (increasing), or flat (constant) relative to the previous position. 3: Mark each turning point — every local maximum (peak) and local minimum (valley) — with its x-coordinate. 4: Record the x-values where the downhill stretches start and stop. 5: Write each downhill stretch as an open interval between consecutive turning points.

If the graph has a break, jump, or vertical asymptote, restart the trace on the other side. A rational function like 1/x is decreasing on (0, ∞) and also on (−∞, 0), but it is not decreasing across the asymptote at x = 0, because the function isn't defined there and the values leap.

Failure handling: if you can't tell whether a segment is flat or slightly sloped, plot two clear points on it and compare their y-values. If the graph shows a plateau, mark it as constant rather than forcing it into increasing or decreasing.

Convert Peaks and Valleys Into Open Interval Notation

Interval notation is where most lost marks happen, and the rule is simple once you see why. Report decreasing intervals with parentheses, never square brackets. The reason: at a turning point, the function is neither increasing nor decreasing, so the endpoint doesn't belong to either interval. For a graph that decreases from x = 1 to x = 4, write (1, 4).

Reading the intervals off the turning points is mechanical. The function decreases between a local maximum and the next local minimum. Find the peak's x-value, find the next valley's x-value, and bracket them with parentheses. If there are two separate downhill stretches, list both, separated by a comma or the word "and."

Some curricula, including several AP Calculus course descriptions, accept or even prefer closed endpoints on monotonicity intervals because the function is still non-increasing there. Your safest move is to match the convention your teacher or exam board uses, and to state it once in your work. When in doubt, open intervals are the standard default in most algebra and precalculus courses.

Use Turning Points to Split the Domain Into Test Strips

A graph with several bends becomes manageable when you slice the domain at every turning point and asymptote. Each slice is a "test strip" with a single direction of travel. On a smooth polynomial, alternating peaks and valleys guarantee the directions alternate too: down, up, down, up.

Take f(x) = x³ − 3x, which has a local maximum at x = −1 and a local minimum at x = 1. The turning points cut the number line into three strips: (−∞, −1), (−1, 1), and (1, ∞). The middle strip runs downhill, so the decreasing interval is (−1, 1). The outer strips run uphill and belong to the increasing answer.

This strip method also handles quartics and higher-degree curves. Count the turning points, and you know the maximum number of monotonic pieces. A quartic has at most three turning points, so at most four strips. If you find a repeated root or a flat inflection, a "turning point" may not actually reverse direction — check the slope on both sides before assigning a direction.

One edge case deserves a flag: an inflection point where the curve flattens but keeps descending, such as the middle of f(x) = −x³, doesn't split the decreasing interval. The whole real line stays decreasing, because the direction never reverses.

Steepness Tells You How Fast, Never Whether

A graph plunging steeply on the left can fool you into calling the steepest part "most decreasing" and the gentle part "not really decreasing." Steepness describes rate, and rate doesn't decide direction. A gentle downhill grade is still downhill.

The derivative makes this precise. The value of f′(x) is the slope of the tangent line, and its sign gives direction: negative means decreasing, positive means increasing, zero means a horizontal tangent. The magnitude of f′(x) tells you how fast the change happens. A slope of −5 and a slope of −0.2 both sit on decreasing intervals.

This matters on real graphs with varying curvature. On a parabola opening upward, the left branch descends steeply near the vertex and then levels off as it extends left. Every point on that branch has negative slope, so the entire left branch is decreasing — the flattening doesn't end the interval. Students who stop the interval where the curve "looks flat" cut off valid x-values.

For the calculus-level answer, solve f′(x) < 0 and confirm against the graph. When the algebra and the picture disagree, trust the algebra and re-examine your reading of the graph, because hand-drawn or low-resolution plots can hide small features.

Fill in Vertical Asymptotes and Domain Gaps First

Discontinuities are the most common reason a correct-looking answer is wrong. Any x-value where the function is undefined — a hole, a jump, or a vertical asymptote — must be removed from the domain before you assign intervals. Drawing the trace right across an asymptote produces a false "decreasing everywhere" claim.

Consider f(x) = 1/x. Its graph falls continuously on both sides of the y-axis, tempting a student to write "decreasing on (−∞, 0) ∪ (0, ∞)" as one merged interval. That's incorrect. Across x = 0 the function jumps from very negative to very positive, so the left-to-right comparison rule fails between points on opposite sides. The correct statement lists two separate intervals.

Treat each piece the same way. For a step function, a piecewise definition, or a graph with a removable hole at x = 2, split the domain at the gap. Then run the pencil sweep on each piece independently and report the intervals piece by piece.

Check the domain edges too. If the graph is drawn only for x between 0 and 6, your answer can't extend past those bounds. The decreasing interval lives inside the drawn domain, not beyond it.

Compare Open Versus Closed Endpoints Before You Write the Answer

Endpoint convention is a real source of point loss because different courses grade differently, and the underlying math supports more than one honest choice.

The open-interval case rests on the definition: at the turning point itself the function is neither increasing nor decreasing, so the point is excluded. The closed-interval case rests on a weaker condition (non-increasing), which holds at the endpoints. Both are defensible. Your job is to know which one your grader wants.

Practical rule: default to open intervals unless your syllabus says otherwise, and if you have room, add a one-line note such as "decreasing on (−1, 1); endpoints included or excluded per course convention." That note costs nothing and protects you either way.

Route the Interval to the Right Test: Precalculus Versus Calculus

The graph-reading method works at every level, but the expected supporting work changes as you advance. Knowing which tool your course wants keeps your answer aligned with the grading rubric.

In precalculus and algebra, you read the graph directly. You identify turning points visually, state the intervals, and justify them by describing the downhill travel. Some courses ask for a table of sampled points as evidence.

In calculus, you compute. Find the derivative, solve f′(x) = 0 to locate critical points, build a sign chart, and report where the derivative is negative. The graph serves as confirmation, not as the primary tool. AP Calculus exams routinely ask for justification, and "because the graph goes down" won't earn full credit there.

If your course uses a graphing calculator, the built-in minimum and maximum finders give you turning-point x-values to several decimals. Use them to check your visual reading, but still write the interval from the turning points rather than from the raw calculator output, since the decimal is approximate.

Diagnose False Decreasing Claims From Misread Graphs

Real graphs mislead, and a few repeat offenders account for most wrong answers. Recognizing them turns a shaky read into a reliable one.

The top error is reading the y-intercept as a turning point. A curve crossing the y-axis isn't necessarily changing direction there. Check the slope on both sides before splitting the interval. Another is treating a steep drop and a subsequent slowdown as two different directions; they're one direction at two rates.

Scale distortion is a quieter problem. A graph stretched vertically can make a mild increase look steep and a real decrease look flat. Check the axis labels and tick spacing before trusting your visual read, and when you can, compare two plotted points on the segment to confirm the sign of the change.

Finally, watch for overlapping intervals or a missing piece. If a function decreases on (0, 2) and again on (2, 5), a combined answer of (0, 5) is only correct if the middle point genuinely continues the descent, which a turning point at x = 2 would rule out. Recheck each junction.

FAQ

Is a function decreasing if its graph is below the x-axis?

No, position relative to the x-axis says nothing about direction. A function can sit entirely below the x-axis and still be increasing — f(x) = x − 10 is increasing everywhere despite negative y-values for x below 10. Direction comes from comparing consecutive x-values, not from the height of the curve.

Can a function be both increasing and decreasing at a single point?

No, because monotonicity is a property of an interval, not of an isolated point. At a turning point the function is momentarily flat, and the tangent slope is zero, so the point belongs to neither the increasing nor the decreasing interval. That's the exact reason decreasing intervals use open endpoints.

What is the decreasing interval of f(x) = x² on its full domain?

The parabola decreases on (−∞, 0). Its local minimum sits at x = 0, where the curve bottoms out, and the right branch increases on (0, ∞). If your course wants closed endpoints, the answer becomes (−∞, 0], and you should follow whichever convention your teacher stated.

How do I handle decreasing intervals when the graph has a hole?

Split the domain at the hole and report separate intervals on each side, exactly as you would for a vertical asymptote. A hole means the function is undefined there, so a single interval spanning the gap would compare values across a point that doesn't exist, which violates the definition.

Do increasing and decreasing intervals ever share an endpoint?

Only when the function is constant between two turning points or when a course counts closed endpoints. On a strict reading, the turning point itself separates the intervals and belongs to neither, so the intervals sit side by side with a gap of one x-value. Constant stretches become their own category rather than joining either neighbor.

How many decreasing intervals can a degree-five polynomial have?

A degree-five polynomial has at most four turning points, so it can have at most two separate decreasing intervals. The directions alternate across the turning points, giving sequences like down-up-down-up or up-down-up-down. A graph showing three distinct downhill stretches would signal a computation error.

Which is more reliable for this, the graph or the derivative?

The derivative is more reliable when you can compute it, because it gives exact critical points and handles subtle flat spots that a hand-drawn graph hides. The graph is faster and works when no equation is given. Use the graph to build intuition and the derivative to confirm, and trust the algebra when the two disagree.

Does a straight line ever have a decreasing interval that stops partway?

No, a non-vertical straight line has one constant direction on its entire domain. A line with negative slope is decreasing on (−∞, ∞), and a horizontal line is neither increasing nor decreasing. Only curves that change direction produce bounded decreasing intervals.