Assigning polarity to connectors in your model lets you visually indicate the cause-and-effect relationship between model variables. Assigning polarity can also make it easier to tell a story about the feedback loops in your model.
For example, in the following model, when population increases, its effect is to also increase the effect of crowding (note the "+" polarity sign end of the connector). This is a positive polarity. Conversely, as carrying capacity increases, the result is to decrease the effect of crowding, a negative ("-") polarity.
Choose the polarity you want to assign (positive or negative) based on the interaction between the two variables.