It’s central to our conception of ourselves as rational agents that we do the things we do because of what we think. What, then, must thoughts be like in order for that conception to make sense? Among other things, thoughts must be transparent – we must normally be able to know what we’re thinking (without the aid of reasoning). Discussions of the transparency of thought have often been framed within the specific contexts of certain disputes (internalism versus anti-individualism; Fregeans versus Relationists), and I feel this has made people gloss over one important fact: the need for transparency is a precondition of the intelligibility of thought and reasoning as a whole, and is thus not a specific problem about singular thought. Appealing to mental files or non-descriptive modes of representation then cannot be the whole story. My hunch is that we must say that thinkers normally have non-propositional knowledge of their own thoughts – such as acquaintance-based knowledge of knowledge-how. All this suggests that we have a special type of self-knowledge regarding our own occurrent thoughts. One wonders whether this is compatible with another central feature of rationality: the publicity of thought. For example, the explanatory role of thoughts in folk psychology is supposed to be generalisable both intersubjectively and diachronically: thinkers who think alike, behave alike; thinkers who don’t change their mind, repeat their past errors etc. In the last part of the talk, I want to discuss possible ways of accommodating the idea that thoughts aren’t only transparent but also public. Egocentric thoughts will be my case study: how is it that a past perception can be taken as univocal with a current episodic-memory thought? How is that a today-thought from yesterday can be taken as univocal to a yesterday-thought today?