Imagine the scene: you're playing your newest favorite game for the first time; you play through the tutorial and first level, and the game begins to unfold itself to you. You find an item or gadget that doesn't appear to do anything just yet, or see a locked gate that you can't open. You might not know what that item does or where that path leads, but one thing is certain: your frame of reference within the game just got larger. You're suddenly aware of the existence of other bits and pieces of this world, and how, chances are, these items and doors are going to get used and opened (respectively, presumably) sooner or later! The world feels larger for these discoveries, and the game's level designers didn't have to do a bit of extra work to do it.
These techniques and many others strive for what's called worldbuilding, where a game is made to feel less like a series of levels and more like a living, breathing space. While the advent of modern technology has certainly made it possible to create massive, sprawling worlds, for most teams doing so is either prohibitively expensive and time-consuming, or might simply detract from the focus of the game. However, just because a game is smaller or doesn't need a massive open world doesn't mean that it can't feel like a real, navigable space for players. There are a lot of ways to accomplish this, but I'm going to focus on just three in this essay: references, obfuscation, and nonlinear layout.
Likely the easiest of the three, referencing other parts of your game world in early game is a great way to clue the player into the fact that your game is more than a simple progression from level to level. Lots of games do this in very simple ways; receiving a new item to find that it's effective or necessary for progression in another level is a great way to stitch together different areas into a cohesive-feeling whole.
Although Batman: Arkham Asylum gives players the ability to see destructible walls very early on in the game, it doesn't give them the ability to blow them up until later. Getting this ability after already having seen the walls indicates to the player that the world is wider than they previously thought.
This sort of thing gives the feeling that different levels aren't necessary these wholly distinct worlds from one another; if you get a can opener in level B and you saw some some cans earlier in level A, levels A & B start to have some continuity with each other. They start to exist in the same universe, and if you can layer these kinds of references across your levels in a natural way, you've made a big step towards elevating your game from a series of levels into something more.
Another way to use references to aid worldbuilding is to incorporate writing in a way that complements gameplay. References to items, people, or places in the story are a great place to start, but if they foreshadow real, in-game things that players can experience, suddenly the game's writing is justified. Lore documents found scattered around are no longer just a way to keep the writing team busy (and how!), but rather they now represent possible clues to things later in the game. The first time you can "keep your promise" to a player in this way, you build a sort of trust between game and player that writings found in-game can actually come to bear. While this isn't always a good thing (try not to overpromise and underdeliver), it definitely goes a long way.
Bloodborne (and the Souls games in general) reference real locations often in books and item descriptions littered around the world. In this case, Cathedral Ward and Old Yharnam are both places that the player can reach later in the game; by delivering on this "promise," the game's writing is validated and given purpose.
Another effective way to make bigger-feeling worlds in games is to obfuscate the parts that define its structure. Hiding a game's length, number of levels, or really any possible indicators of progression that live within menus and substructures will help with this. A level-select screen is both the biggest indicator that players are in a game (rather than a world) and the biggest tip-off of when a game is about to end.
For example: LittleBigPlanet 3 is interesting because it offers the promise of "hub levels," where players can navigate a space and move in-and-out of distinct story levels. However, in the level select screen, which is shown before players arrive in the hub level (the biggest circle with the gorilla on it), players are shown all of the hub's connected levels. Players are immediately given an idea of the overall scale and possibility space of the upcoming section of game, thereby taking away from the exploration and surprise that the hub levels are meant to provide! It's a strange design choice that likely hearkens from the earlier two games' level-based structure, but it undermines the player's potential for discovery.
This might represent the most difficult of the three techniques described in this post, but it's also possibly the most effective. Designing your game to incorporate levels that circle back on themselves, have multiple paths, or have unexpected shortcuts or connections with other levels, is an excellent way to create a world that feels real.
Your world needn't be as complicated and intertwined as the ones from Castlevania or Metroid, but a key takeaway from this style of game is the multiple entry and exit points in each area. Finding shortcuts that lead to familiar areas builds a sense of understanding of the world as a real, navigable space rather than simply a bunch of rooms.
The real world is rarely so linear as most games, and even hints of nonlinearity in level design can make a level (or even a game) feel broader and most expansive than it perhaps really is. Implementing this is definitely effortful; it's harder to make a branching path than a linear one. However, the sense of surprise and discovery that comes from a player learning that two paths that they had previously explored were actually connected is hard to replicate, and sets a precedent that gives the impression of a genuinely bigger world.
Making use of players' curiosity and surprise through reference, obfuscation, and non-linearity can make for compelling and exciting worlds in otherwise smaller games. Although it can be challenging to effectively implement these techniques, and although they often demand extra time and effort, the net effect can be magical, resulting in not just a game, but something closer to a world.