People who on waking tell us certain incidents (that they have been
in such-and-such places, etc.). Then we teach them the expression
"I dreamt", which precedes the narrative. Afterwards I sometimes
ask them "did you dream anything last night?" and am answered
yes or no, sometimes with an account of a dream, sometimes not. That
is the language-game. (I have assumed here that I do not dream myself.
But then, nor do I ever have the feeling of an invisible presence;
other people do, and I can question them about their experiences.)
Now must I make some assumption about whether people are
deceived by their memories or not; whether they really had these
images while they slept, or whether it merely seems so to them on
waking? And what meaning has this question?—And what interest?
Do we ever ask ourselves this when someone is telling us his dream?
And if not—is it because we are sure his memory won't have deceived
him? (And suppose it were a man with a quite specially bad
memory?—)
Does this mean that it is nonsense ever to raise the question whether
dreams really take place during sleep, or are a memory phenomenon
of the awakened? It will turn on the use of the question.
"The mind seems able to give a word meaning"—isn't this as if I
were to say "The carbon atoms in benzene seem to lie at the corners of
a hexagon"? But this is not something that seems to be so; it is a
picture.
The evolution of the higher animals and of man, and the awakening
of consciousness at a particular level. The picture is something like
this: Though the ether is filled with vibrations the world is dark.
But one day man opens his seeing eye, and there is light.
What this language primarily describes is a picture. What is to be
done with the picture, how it is to be used, is still obscure. Quite
clearly, however, it must be explored if we want to understand the
sense of what we are saying. But the picture seems to spare us
this work: it already points to a particular use. This is how it takes
us in.
Pictures - This is how I think of W's use - throughout PI, of pictures.
I will give a concrete example of what I think he is stating in very general, abstract ways.
In logic, a formal language and a formal system are supposed to be sets of 'meaningless' symbols
along with rules of syntax and transformation. In sentence calculus, everyone is familiar with the
variables - p.q.r - and the connectives - NOT &, etc.; as well as what order in which they are allowed to be
placed next to one another (syntax) to form formulas; and then the rules of transformation such as substitution and detachment (modus ponens).
Logicians like to say that these symbols are meaningless until they are interpreted, or given a model. However, this
may be a good pedantic tool to show students that formal systems can have multiple models; and for showing the distinction
between formal system and interpretations - but to say that the formal language is completely separate from its intended interpretation goes a step too far. One could not set up the axioms and syntax of a system without reference to the intended interpretation.
For example, Hilbert famously said, illustrating the above concept, that he could just as well use symbols such as 'beer mug',
'chair' and 'table' rather than 'point', 'line' and 'plane' in doing Euclidean geometry. However, in truth, he could not have set up the axioms and rules with such symbols - there is no way one would think to say that 'two beer mugs (points) uniquely determined one 'chair' (straight line).' The axioms are determined by (set up according to) the intended interpretation. It is only after the axiom system is set up that the intended interpretation may be discarded. Logicians do recognize this, but the point is often missed. In this way, formal systems are 'guided' by a pre-existing model or interpretation and we must look within that model (picture) to find the underlying assumptions which are then expressed by axioms.
I think the above is a special case of what W generally means when he speaks of pictures as in this the quoted entry. We work within models/pictures when we think and talk. The models supply the framework for language games (systems). The unstated assumptions we utilize everyday - such as treating people as 'not automatons' or that dreams are not conscious memory mistakes - are not only unstated, but subconscious as well. They are part of the model in which we are working and we must look at the model to find them and make sense of them. In this way, the picture precedes the discourse in a way - meaning is supplied by the overall picture, not sentence by sentence.
How humans think and talk in such a way as to follow a picture or model even prior to articulating or formalizing the
picture is a great mystery and at the heart of rational thinking.