Science requires falsifiable propositions, preferably assembled in structures that provide nonobvious predictions across a wide scope. The human behavior and probability branches of fire safety science meet this definition. The topic of fire safety science, as distinct from fire science, demands human behavior and probabilistic modeling. In both human behavior and probabilistic modeling applied to fire safety science, there are multiple schools and approaches, often reflecting disputes over how much it is necessary or possible to imitate the structures and elements of the hard branches of fire safety science. A review of these approaches and the state of the art provides a basis for greater understanding of the nature and importance of the soft branches of fire safety science -- and the hard fact that they are worthy of the name "science" and essential to any valid and useful fire safety science structure.