Humans (and probably anything else) don't have access to the complete set of raw data that makes up our environment. We have imperfect data streams coming from our sense organs and from that we build a model of reality, which may have varying levels of accuracy for different people (e.g. we could say a schizophrenic has a disturbed model of reality). The idea being that reading a lot books, traveling, or in general exposing oneself to as much varied data as possible will improve the accuracy of your model of reality.
Have been reading and researching for my postgrad and in some cases "Perspective" & "Mental models" are used interchangeably.
However i like to think of it in this way:
Perspective: The point of view from which something is looked at
Mental Model: A mechanism or thought pattern to make sense of something or understand something about some situation
You can never break out of your perspective of looking at the world. So a mental model is always confined within a specific perspective.
Example:
Supply&Demand is a model that helps you understand pricing in markets. The point of view/perspective is that of an economics/economist perspective. You could just as easily look at the same situation from a different perspective such as sociology or even psychology.
In other words, mental models are like a short cut to look at situations, solve problems and understand the world from different perspectives.
You could say that humans understand the world by building a "model" of it in our minds. When we are trying to decide how to act, we can simulate a situation by running it through the model. For example, you might have a model of your boss that helps you predict how they will react if you give them bad news. And you might have a mental model of your country that you can use to predict what will happen if it goes to war or goes broke. Every time you learn something, you can use it to adjust your mental model. It's like a simulation of the world inside your mind. Pretty cool, actually.
People can end up with very different mental models of the same objects because they trained them with different data (ie, had different experiences growing up) or because other mental models impacted the interpretation of the data, and the development of those different mental models.
It refers to a person's perception of the outside world and how well the person is able to predict the future and understand the causes of events. Sort of like how a machine learning "model" is the mechanism by which a program understands it's inputs.