How to Interpret Neuroscience Figures

Begin with the figure's job

Before reading individual points or colours, look at the figure number, title and caption. Ask what question this visual is meant to address. A figure may compare two task conditions, show a time course, locate a signal, relate two variables or document the steps of an experiment. Its job determines what a meaningful pattern would look like.

Read the caption once without trying to decode every detail. Identify the participants, animals, cells, recordings or model involved; the conditions being compared; the outcome being shown; and any symbols or abbreviations. Then locate the relevant sentence in the results section. The text often states the authors' intended claim, while the visual lets you check how directly the displayed evidence supports it.

Describe before you explain

Make a plain observation first. For example: “The average line rises after the cue in one condition and remains flatter in the other.” That is different from saying the cue improved attention or activated a particular process. The latter statements may be reasonable interpretations, but they require the design, measures and controls to justify them. Keeping these steps separate is one of the most useful habits in scientific reading.

Read every axis and encoding

For a graph, start with the axes. The horizontal axis usually represents categories, experimental conditions or time, while the vertical axis represents a measured outcome. Do not assume this pattern; read the labels and units. A value in milliseconds, percentage correct, fluorescence intensity or a model-derived score has a different meaning and a different scale of interpretation.

Next identify how groups are encoded. Lines, bar colours, point shapes, panels and line styles may each represent a condition, population, brain region or analysis. Find the legend before comparing them. In a multi-panel figure, note whether every panel uses the same scale. A small-looking change may be large on one axis range and modest on another.

Check the reference point

Many neuroscience figures are baseline-corrected, normalized or expressed as a difference from a reference condition. A vertical axis centred at zero may show change relative to baseline, not an absence of activity. A percentage may be scaled to each participant's own starting value, not a raw group score. The caption or methods should explain the reference. Without it, direction alone can be misleading.

Also look for transformed scales. A logarithmic axis changes by multiplicative steps, and a colour scale can compress or expand apparent differences. These choices are often appropriate, but they change what visual distance means. Name the scale before using the apparent spacing of marks as evidence.

Find the unit of observation

When individual observations are visible, inspect their spread as well as the average. Do most values move in the same direction? Are there clusters, outliers or wide differences between participants? A group mean can be informative, but it can also hide variation that matters for the question.

When only summary bars or lines appear, do not infer that every individual showed the same pattern. Look for a sample size in the caption, labels or methods, and note whether the display includes individual values, confidence intervals or another uncertainty indicator.

Interpret error bars with care

Error bars signal that an estimate has uncertainty or variation, but they do not have one universal meaning. Depending on the study, they may show a standard deviation, standard error, confidence interval, credible interval or another quantity. The caption should say which. Their size alone cannot tell you whether a result is important, reproducible or statistically distinguishable from another result.

Ask about magnitude, not just direction

After identifying a difference, ask how large it is in the units that matter. A tiny change can appear dramatic if the vertical axis is tightly cropped. A modest-looking difference can matter if the measure has a meaningful real-world scale. Look for the numerical axis, reported estimates and the study context before deciding whether a pattern is substantial.

Approach brain images as maps with rules

Neuroscience visuals often use brain slices, surface maps, microscopy images or heat maps. These images are compelling because they appear to show where something happens. Begin by finding the colour bar. It tells you what colours encode: a signal strength, statistical value, correlation, classification score or another quantity. Warm colours do not automatically mean “more brain activity,” and cool colours do not automatically mean “less.”

Then identify the spatial reference. Is the image a single participant, an average, a selected slice, a surface projection or a schematic? What orientation is used? A left-right label, slice position or coordinate system can prevent a basic location error. An image may be thresholded so that only values meeting a stated criterion are displayed; uncoloured areas are not necessarily evidence of no effect.

Be cautious about reverse inference: observing a pattern in one brain area does not, by itself, establish one particular mental state or function. Brain regions participate in multiple processes, and the strength of an inference depends on the task, comparison and prior evidence. A careful reading says what signal was associated with the stated condition, then asks what the design permits researchers to conclude.

Read time-course and connectivity displays

Time-course figures show how a measure changes across time. Locate the event markers first: stimulus onset, response, trial phase or another reference. Check whether time is aligned to the same event for every trial and whether the line has been smoothed or averaged. A peak after an event demonstrates timing in the displayed measure; it does not automatically establish a causal chain.

Connectivity matrices and network diagrams need a similar pause. A coloured square or connecting line may represent a correlation, coherence measure, model parameter or structural connection. The legend, threshold and directionality matter. A connection in such a visual may describe statistical association rather than a direct anatomical pathway or a flow of information. Look for the definition before using everyday language such as “communicates with” or “controls.”

Eye-tracking studies provide a useful reminder that a display depends on its measurement. A heat map can summarize where people looked, while a scanpath shows sequence and a fixation-duration plot shows time. NeuGeneration’s guide to eye tracking in neuroscience explains what these measures can and cannot represent.

Use the methods to test the visual claim

A figure becomes more meaningful when you connect it to the study design. Ask what was manipulated or observed, what control condition provides a comparison and how the outcome was measured. If a graph compares two groups that already differed in several ways, its pattern may show an association rather than the effect of one cause. If a measure is indirect, the conclusion should remain proportionate to what that measure captures.

Notice processing decisions that could shape the display: exclusion rules, averaging, filtering, region selection, normalization and thresholds. These are not automatically problems. They are choices that help turn complex data into a readable visual, and a transparent paper explains them well enough for a reader to understand the resulting figure.

For a broader method of tracing a study from question to conclusion, see How to Read a Neuroscience Research Paper. The figure is one important part of that appraisal, rather than a substitute for the full argument.

Build a short figure-reading routine

Use the same sequence each time you encounter a new visual:

1. State the figure's question in your own words.
2. Name the data type, unit of observation and comparison.
3. Read axes, scales, legend, panels and caption.
4. Describe the visible pattern without explaining it yet.
5. Identify uncertainty, variation and any processing or thresholding notes.
6. Compare the claim in the text with what the visual directly shows.
7. Write one remaining question about the design or interpretation.

Frequently asked questions

Do error bars show whether a result is significant?

Not by themselves. Error bars can represent different quantities, and their overlap is not a general test of a comparison. Read the caption and the reported analysis, then consider the estimate and its uncertainty rather than relying on a visual shortcut.

What does a bright area on a brain scan mean?

It depends on the colour scale and analysis. It may represent a signal, a contrast, a statistical value or another derived quantity. Check the legend, spatial reference and threshold before describing the pattern, and avoid treating colour alone as proof of a mental process.

Why do some graphs show dots over bars?

Dots can show individual observations or estimates, while the bar often summarizes them. Seeing both helps you assess variation and sample structure. First confirm in the caption what one dot represents and what the bar and error marks summarize.

Can I interpret a figure without reading the methods?

You can describe its visual pattern, but methods are needed to judge what the pattern supports. They explain the sample, measure, comparison and processing choices that give the figure its meaning.

How can I practise reading neuroscience figures?

Choose one figure from a course reading and follow the routine above. Write a one-sentence observation, a one-sentence interpretation that the design permits, and one question you would investigate. Repeating this with different data types builds fluency without requiring you to master every technique at once.