16

Should figure captions only describe all of the necessary components (e.g. the meaning of axes, etc.) or can the figure caption also include a brief sentence about the main conclusion drawn from the figure. For example, say the axes are labelled "speed" and "time". One caption might be "Speed of a giraffe in meters per second as a function of time in seconds from when it first starts moving." It could also have the additional sentence "... Note that Speed is increasing with time." For the previous example, this is probably unnecessary, but for more complicated "take home messages" the point of the figure may not be as obvious. So should comments describing the main result behind the figure be only put in the text body or is it OK to put it in the caption as well? I think this might be field dependent.

The justification for wanting to put the main message of a figure in the caption is that it makes the paper easier to skim, as, anecdotedly, I've heard quite a few people say they skim the paper and focus most of their effort on the figures before ever looking at the results.

  • "Speed vs. time" doesn't add any useful information, provided you labeled your axes. In your example, I would go with "Figure 1: Note that speed is increasing with time." I'm not aware of any field that frowns on a caption like that, but if you're worried, you can always check previous volumes of the journal you want to submit to. – user37208 Nov 9 '15 at 6:22
  • 1
    You should definitely point out the point of the message if it is not immediately obvious. – Davidmh Nov 9 '15 at 9:36
  • 4
    @PålGD: If the figure is a photo, what is wrong about making it a bitmap figure? – O. R. Mapper Nov 9 '15 at 11:02
  • 5
    @PålGD: Since when are all figures graphs? What about real-world photographs of test setups or user studies? Screenshots of running software? Photos of prototype devices? Paintings or photographs analyzed or processed by something described in the paper? etc.? As a random example: Count the figures in this paper that are photos and that could just as well be shown as a vector-based graph. – O. R. Mapper Nov 9 '15 at 12:19
  • 1
    The answers (and their votes) you have reflect different cultures. For example, in my field but not necessarily others, papers are skimmed in groups while sipping coffee, focusing on the figures rather than text, and if your figure caption doesn't have a take-home message, we'll be mildly annoyed. So the question is, how are papers read in your field? – user4512 Nov 9 '15 at 16:46
8

There are different schools of thought about this; how much information to include in a figure caption depends largely on personal taste.

Proponents of minimalist figure captions put forth that

  • It needs less space.
  • It says in one brief sentence what the figure is about.
  • Having a long block of text below a figure is ugly.

In contrast, proponents of long figure captions state that

  • People who (if only at first reading) only skim over the paper and just look at figures do not have to search in the text what each figure is about. The very points found within the figure are explained in depth in the caption.
  • When looking at a figure in detail, readers do not have to switch forth and back between the descriptive text and the figure, as a concrete description is right below the figure.

Personally, I consider the points in favour of an extensive figure caption more convincing, and I usually go for those as long as space restriction allow so.

An additional benefit is that you can use the paper text for an abstract description of some aspect of your work, and then use the figure caption for an equally detailed, but more concrete description of the example shown in the figure. This way, you improve comprehensibility of your paper without making it seem redundant or repetitive - as a certain degree of overlap (or even exact repetition) between text body and figure captions is acceptable.

  • +1 for your last paragraph, I had never thought about doing it that way. – WetlabStudent Nov 9 '15 at 13:40
18

A short comment in a caption is ok. But generally captions should stay descriptive of what is drawn in the figure e.g.

Fig. X: estimated airspeed velocity of an unladen European swallow.

They should also ideally not be too redundant with legends although sometimes it helps to say things like "shown at various headwind velocities". The place to draw attention on a specific aspect of the data shown in a figure is the results section.

These are customs and not strictly-enforced rules, nothing bad will happen to you if you take a few liberties. Worst case scenario, you'll have to make a few minor edits during revision.

  • "The place to draw attention on a specific aspect of the data shown in a figure is the results section." - who says that any part of a figure must be a result? – O. R. Mapper Dec 12 '15 at 0:20
14

I strongly advocate for captions that are as self-contained as possible, saying not only what the figure is but also the essence of what the reader should take away from it.

Yes, this introduces some minor degree of redundancy, but I believe this minor inelegance is well worth the benefits that can be obtained. Figures are extremely salient for a reader's attention: there are relatively few of them and they each have a strong visual impact, going through a different set of cognitive pathways than text. Many readers first impression of a paper will thus be formed by its figures, as they flip through a paper deciding whether to read it (and many journals explicitly encourage this with "online previews" that show only the figures of a paper). Even once somebody has decided to read a paper, they may not read all of the sections in depth, but may skim some and read others. Finally, even for a reader going in depth, the cognitive availability of the information associated with a figure is increased when the reader can pull key points right out of the caption rather than needing to go to some more distance part of the text.

As such, I believe that insofar as is possible, figures and their captions should provide a complete sketch of the paper narrative. Doing this will significantly increase the impact of your papers by helping readers at many different levels of commitment to reading.

2

It's always worth getting someone to proofread your submissions. All co-authors should anyway, and those less involved with that aspect of the work can be best qualified to judge the clarity of your exaplanations (they can also be most liekly to skim over that section). Otherwise, you have peers, help each other. The best style partly depends where your figure is going.

In a journal paper, styles vary between journals, even from the same publisher. Some like dense figures with long captions, or at least, these become the norm when page limits are tight. Other journals prefer more spacious figures, fewer insets, and simpler captions. The correlation between cpation and figure styles is only meant to be indicative; I'm sure you could find examples of simple figures and very full captions. Some journals prefer the legend to be given in the form of text in the caption.

In a thesis you'll normally have a list of figures, with short captions in it. These short captions should give enough information for the reader to find a figure ("There was a figure relating bandgap and lattice spacing, now which figure was it?") while including a subset of the information in a full caption. Thus the full caption will end up reasonably descriptive.

There are a few other, competing, factors:

  • Help your reader. This is a point I often make, but here could be interpreted as:
    • Don't waste space and the reader's attention stating the obvious...
    • ... but add clarity.
    • If a figure really works only in colour (and these should be rare), provide enough information so the reader knows this, and doesn't juyst hink your figure is rubbish.
    • Consider a reader who may not see too well -- give them the information in text to decide whether to print a large copy. (This is an example, but helping this hypothetical reader may help all your readers).
  • Help the reviewer. This overalps with helping the reader, but a reviewer will be looking for different things. By all means say "for a full description of the samples, see main text", but rather than just "Sample A (red squares), Sample B (black circles)" give them a gentle reminder of which sample is which (this of course is easy if it's a treated and a control population).

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.