Deception in research involves lying to or intentionally misleading subjects. Withholding information may or may not be deception. For example, telling subjects that an urn contains an unspecified number of red and blue balls, and inviting them to participate in an experiment by guessing the number, would not involve deception because the subjects "know what they don't know". Concealing the fact that there is a hidden camera would be deceptive. The distinction is whether subjects have the information to make an informed choice about whether to participate in the research. Secretly filming their behavior denies them the opportunity to consider whether or not they want to be filmed. Not telling them how many balls are in the urn is part of the accepted framework for a participant to guess the number.
It may be important in some research to withhold the specific theoretical purpose of the research from subjects, in order not to bias their opinions. If done in a neutral way, it would not be deceptive. If subjects are intentionally led to believe that the research is for a purpose different than the actual purpose, this would be deceptive.
Deception poses ethical problems and should be dealt with by weighing the benefits of the research against the harm (if any) from the deception. Aside from any potential harm to participants, deception can also harm the institution by building the perception among potential subjects that "researchers are liars."
If deception is used there should be a planned debriefing after the research is completed and this debriefing should be addressed in the protocol and informed consent (if applicable) requiring IRB review and approval.