The need of animals to visually recognise things in their environment for what they are, and in the case of humans - face recognition, are key survival skills. This cognitive ability has evolved in humans well before language was invented, let alone written one.
As such, the brain ability to recognise imagery is quicker than its ability to interpret written words.
Accordingly, the visuospatial memory is the strongest, fastest and largest of all other types of memories (albeit some people are better utilising other types of memory, but this ability is acquired, not inherited).
What's more, when it comes to face recognition, despite their perceived complexity, the set of visual generalisations the brain makes is marvellously basic and simple - face recognition happens in an instant.
Serial vs parallel
Most people's names have less than 3 words, which means the eye can scan each name with a single fixation - this is good for visual searching. However, text is read using eye fixations in a serial manner; this means that in order to read a shopping list (or a list of contacts), each item has to be scanned - one at a time (this is true if you actually reading it; for visual search the brain takes shortcuts, but it's a process that with words still involves more fixations).
With a group of images, a single fixation could suffice as in addition to the point of fixation the brain also processes in parallel the peripheral view.
For example, if I'd ask you to find the Kiwi in the next image, the priming for green and oval shape mean you'll be able to spot it with a single fixation (scroll down quickly to see this in action - you should spot the kiwi without even trying to).
To demonstrate the feature-generalisation nature of visual cognition, look at the following image from far (so the fixation is on the whole matrix, not a particular cell within it). You'd still be able to spot with ease the apple.
While you are also primed when searching for someone's name, your brain still has to utilise more fixations and the process is largely serial. Find the pear:
- Blood orange
Did your eye briefly fixated on the Papaya? Can you think why?
From personal experience
There is a usability issue with Spotify on mobile that relates to exactly this.
In my music library, I have many playlists. The playlist list involves images of the cover:
When I look for a particular playlist, I scroll through this list real quick as the search is for imagery cues.
However, for a playlist itself, Spotify has no cover images:
This makes sense if the whole playlist is of a single album. But I have a playlist of 50 different songs from 50 different albums, and trying to find a song in this list is really difficult compared to finding a playlist in the previous view. (And I do this every morning, with the same playlist, and it's a real pain.)
Adding avatars will decrease the time it takes people to recognise the person they're after. Using a grid rather than list view, when possible, will take advantage of the brains parallel visual processing for peripheral view.