The wideband sensors use an internal chamber that is heated to about 1200F, with a platinum catalyst to finish reacting any unburnt fuel/air mixture. What is left in the chamber is either extra oxygen or extra fuel.
Then, it uses a bi-directional electro-chemical oxygen pump to either pump external oxygen in or out of the chamber to maintain a 450mv reading on its internal NB sensor.
The current, positive or negative, applied to the pump circuit determines how much oxygen was pumped in out out of the test chamber to maintain the lambda 1.0 ( 450mv ).
If it had to pump oxygen IN to the chamber, the mixture was rich. It if had to pump oxygen OUT of the chamber, the mixture was lean. The pump current is converted to o2% / lambda / afr.
That's why they can measure lean or rich mixtures. And, why wideband sensors need controllers, rather than just reading a voltage off of them, like a NB.
Also, in reality, a NB sensor can only really sense rich conditions. It measures the oxygen content of the exhaust gas vs. the atmosphere. If the difference is greater than some preset limit ( usually near the partial pressure of oxygen in the atmosphere STP ), it generates a voltage, based on the difference on either side of the sensor ( exhaust or free-air ). The curve is very steep. And generates about 1.0v at the maximum difference. Pending the sensor temperature, atmospheric pressure, and exhaust gas pressure. That's why they try to keep it at around 450mv. In the middle of its useful range.
As far as trusting them; that is a different story. It depends on the health of the sensor, the controller, and how it is connected to the logging device.
And that is a thread, all on its own