In Althusser's celebrated notes on ideology, the distinction was made between Repressive State Apparatuses and Ideological State Apparatuses. The former (hereafter RSAs and ISAs) refer to those departments of the state that back its authority up with force. They are the "armed bodies of men", as Engels put it, and covers the police, the military, the secret services, the penitential system and, you might argue, paramilitaries. Their primary purpose is to repress. Of course, the size of these apparatuses and their "activism" vis a vis the general populace varies from country to country. In a liberal democracy these institutions are much more in the background, whereas the grotesquery of something like North Korea relies on an overblown repressive apparatus to keep the despotic Kim dynasty in power.
Complementing this for Althusser are a set of apparatuses of a different character. An ISA is an institution that secretes ideology as its primary purpose. These include schools, the media, the family, religious denominations, all forms of popular culture, and so on. While they're not all formally part of the state, they remain a state apparatus because the ideas, values, and normative practices they promote work to sustain bourgeois state power. This is an understanding of ideology shared by Poulantzas, who notes that "ideology does not consist merely in a system of ideas or representations [which is very much the mainstream bourgeois view - PBC]: it also involves a set of material practices, embracing the customs and life-style of the agents and setting like cement in the totality of social (including political and economic) practices." The state also helps organise ideology and deploys it for its purposes "... to legitimise violence and contribute to a consensus of these classes and fractions which are dominated from the point of view of political power." (p.28) What also made Althusser's discussion of ideology important was his attempt to how ideology constituted (interpellated) the members of society as subjects capable of functioning in them.
Poulantzas extends the interpellation argument. As ideologies generate subjectivities, they do so by constraining, manipulating, and consuming human bodies and in so doing the body becomes a political battleground. This positioning of the body comes about via two routes. There are institutions that discipline bodies as agents pass through them - think how RSAs and ISAs alike regiment their charges. And there is a more diffuse 'bodily order' that permeates the cultures of advanced capitalist societies, a set of discourses and ideologies that bend and mould bodies so they're suitable for entry into a variety of apparatuses. This is not an accidental by-product arising spontaneously from the bowels of capitalism either. The disciplinary institutions and their discourses require a cadre of professionals who refine and elaborate these procedures, and it happens that the majority of this milieu are employed by the state. Hence, Poulantzas argued that the state is creative and reproductive of its own authority. The political and cultural consensus is always an accomplishment, and one that is rooted in definite processes and relationships amenable to materialist analysis.
Poulantzas is also adamant that we must not assume that the bodily order and ideology generally is uniform a la the kinds of critique associated with the Frankfurt School. Ideology is always composed of several discourses designed by different apparatuses to be received by certain classes and class fractions. Think about how ideologies surrounding the military address recruits to the ranks and officerdom respectively, position and discipline them differently, and how this in turn differs from the images projected at the public-at-large. Poulantzas is also keen to get away from the simplistic notion that ideology is always a matter of concealment, a conception rooted in perspectives dependent on so-called 'false class consciousness', which in turn is wrongly attributed to Marx. Poulantzas argues that some discourses generated by the apparatus are addressed to dominant classes and their allies and seek to organise their interests. The intention and sometimes the content of these discourses go unconcealed because they have to be quite explicit and tell the truth (or, rather, their truth) to themselves. The pages of the FT, discourses around terror and policing, even arguments within the Tory party often assume this character: all seek to unify a plurality of their class around their interests.
In one respect developing Althusser's views on ideology (which also allows Poulantzas to consider Foucault's approach to discipline later on) pushes them in a more practical direction than the level of rarefied abstraction Althusser's musings suggest. But in going beyond them, he has to also break with the dichotomy between RSAs and ISAs. While a useful way into thinking about the state as an active agent that secures its own material reproduction by churning out ideology and organising the interests it stands for, Althusser's rush to break with instrumentalism and economism completely diminishes the state's roots in the economy, or the very social relations that make the state a bourgeois state. The problem for Althusser is that by boiling RSAs and ISAs down to particular functions, he runs the risk of treating them in an essentialist manner, as expressions of those functions. And if that wasn't bad enough, the evacuation of the economy from his treatment of the state leaves them as free floating institutions tethered only by their functions and not by their place in the constellation of class relationships, founded in an economy that tends to reproduce them. The key task for socialist theory is to map and understand this complexity so it can be changed. And on this score, as far as Poulantzas was concerned Althusser's views were not a helpful contribution.
More Poulantzas here.