Please advise which of the following approaches is more usable:
Switching images, all captions are visible:
All images and captions are visible:
Swithing captions, all images are visible:
The important question about which option is best for your users is "What information do they need in order to choose?"
For example, if the image is a supporing icon to the text, you would pick option 1. If it's a 'choose your favorite picture', you'd pick option 3. And a political cartoon with a text punchline would use option 2, since both parts are critical.
I would definitely go with the third option. It is the most intuitive and it flows much better than the second one (we read right to left). The problem with the first one is that it will rely on the users memory to make the best choice and they will have to scroll through them a few times to make the choice they want to.
If the choice is amongst the three go for 3rd option (assuming image click is workable in your case). It would be fancy way for a user to choose image (answer) rather then old fashioned radio button. you will also save some good space and explanation.
You can also put the description in image title (but, not a good approach to show data in this way , though) for telling user what that image is for.
I think without real images or context, it is difficult to say what is the best approach. When the images you show are all alike, the text is some generic description, and no background of scope, don't expect people to give the best answer to you.
I've had real life situations where radio buttons with only text and no images worked better; on other situations I had push like radio buttons (like the 4 Text Alignment Buttons of the MS Word toolbar) with a text tooltip making the best UI.
As Steve Krug says: "It depends!"