In the Philosophical Investigations, Wittgenstein queries whether facial recognition always involves having a mental image of the currently perceived facial expression. Is this experience typical in cases of facial recognition? It seems atypical. In some cases no mental image may occur at all. Or, a mental image occurs only after one has recognized a facial expression in another’s face. Imagine arriving at a class reunion. One may meet an old friend there, and directly see her in one’s mind as she was twenty years ago. Or, you recognize expressions of gloom and boredom on the faces of the participants, but suddenly you can’t help thinking of the fun everyone had at your family get-together last Christmas. In this last situation, not only do mental images follow recognition of the facial expressions, they are of something else associated with the current impression. Notwithstanding the fact that we sometimes have memory-images when we recognize a face, that fact is exceptional, and hence cannot be sufficient to account for recognition in a general way.  

This brings us to the memory-image itself. According to the philosophers, in recognition one compares one’s present impression with a stored memory-image. The memory-image is treated as a picture or photograph. Is a memory-image like a picture? One may say that it is misconceived to compare a mental image with a picture. A memory-image informs me that it is an old friend I currently see only if I remember that this is what so-and-so looked like. I do the remembering, not the memory-image. The memory-image can only remind me of what so-and-so looked like, and its occurrence presupposes memory and recognition. Russell suggests that a special feeling of familiarity connects a memory-image with the past; justifies it as a symbol of a prior experience. But I recognize the special feeling by remembering it, not the other way round. The special feeling is not recognized independently of memory, and is not evidence for how things were. It takes memory to inform us that what is presented is a representation of the past (Wittgenstein, Zettel, §662).  

The inclination to view individual recognition after the event may betoken contrasting it with matching a picture to what it represents. Ordinarily, when we recognize someone’s face we don’t compare recognition with fitting a picture with what it represents, but in philosophy we routinely do this. Now, there is nothing wrong with this per se. Philosophers look at something when no one else ordinarily looks at it, and in such a way that often generates puzzlement or doubt. If so, then we should remember that it is possible to look at recognition in at least two ways: we can look at the act of recognition after the event to see what actually happened; we can also look at it by presuming that ordinarily my current impression of a face coincides with an imprint in my mind, or that I compare my current impression of someone’s face with a memory-image, and so on. According to Wittgenstein, what appears inevitable in a philosophy is often the preconceived idea we bring to the table and which we never learn to question. Again, philosophers should be mindful that it is atypical that people have memory-images when they recognize someone’s face. This important fact should not be forgotten 

Wittgenstein does not deny that we sometimes have memory-images when we recognize someone, or when we recognize a facial expression. What he does deny is that having such an image is necessary or sufficient for the truth of ‘A recognizes x’. As we have seen, I may recognize x without any mental image occurring to me, and I may have a memory-image of x without thereby recognizing x.