A while ago, I spoke to an investment banker who believed that he was far too clever to have unconscious biases. His argument was that he was so intelligent, everything passed through his conscious mind in order to get into his unconscious mind. He therefore believed it was impossible for him to have any unconscious biases at all, and moreover that his unconscious was largely redundant.

I sincerely hope that is not the case, given that our conscious mind only makes up around 10% of our brain’s thinking and reasoning capacity – to have 90% of our capacity sitting there unused seems like such a waste.

His perspective though does give us an interesting insight into how, despite all the coverage that there has been over the past decade around unconscious biases, they can be very hard to recognise, and harder still to tackle.

One of the most fascinating areas of my work as a diversity and inclusion specialist is conducting live bias reviews, in which I join key decision making meetings client side with the sole purpose of identifying which types of bias are influencing the discussion and decisions.

I have sat in on recruitment wash-up discussions, succession planning meetings, bonus and appraisal-rating moderation meetings as well as bonus calibration discussions.

During the discussion I am looking out for biases introduced by the process, by the people in the room, as well as the conditions under which the decisions are being made.

In excess of 170 forms of unconscious bias have now been documented, although truth be told some of those are simply renamed variations of others. Of this long list of biases to keep an eye out for during these reviews, time and again there are four biases that are most common.

The quotes below have been taken from live bias observations.

Biases introduced by individuals in the room

1. Descriptive stereotypes

One of the most intriguing biases that exists is the prescriptive stereotype. In essence, these are mental descriptions and images we carry in our heads of how men and women should behave – how we expect them to behave.

For example, women are expected to be nurturing, communal and modest. These descriptive stereotypes can be a double-bind for women in promotion discussions, for example, as they can be marked down for not being very confident, and similarly marked down for being overly-confident.

“She’s holding herself back. Would we always be having to drag her to the fore?”

“There’s something about the tone of her application – there’s no humility.”

There is an invisible, seemingly impossible tightrope to be traversed here by ambitious women.

2. Framing effect

This is one of the more insidious types of cognitive bias that affects our ability to make objective decisions. In these key meetings, the decision makers around the table are likely to react to a particular choice depending on how it is presented.

When discussing candidates that are borderline for promotion, for example, we see that line managers are more likely to frame the discussion for minority employees in terms of the risk that their promotion would bring to the organisation:

“She’s not quite the finished article, and she’s not expecting promotion this year anyway.  Let’s tell her one more year, and see what she can achieve in that time.”

“She looks like a leader for the future, but not now.”

“His business case [for promotion] hasn’t come through as strongly in his form as it could have done, but I think this next role could be the making of him.”

This contrasts directly in the same meeting when borderline majority candidates are discussed. Here, the stretch offered by this promotion is instead positioned as a great development opportunity. Ultimately this means that some employees have a higher informal bar than their majority counterparts.

Neuroimaging also tells us that framing effect is specifically associated with amygdala activity. This indicates that our emotional system plays a key role in mediating this type of decision making process.

Biases introduced by the conditions of the meeting

1. Under pressure

It is not uncommon in these meetings for people to have 90 minutes to discuss 30, or sometimes more, employees. All too often, very little thought is given to how time might be best managed. This issue is then further compounded by the fact that people then enjoy a leisurely 20-30 minutes talking about the high performers that everyone agrees on, leaving an ever squeezed amount of time to cover the borderline employees. This is especially problematic as most bias is observed for employees in these borderline groups, and when decisions are made under time pressure.

2. Advocates and sponsors

This one is an old chestnut, but still its impact is profound. Those who have advocates or sponsors in the room fair significantly better than those who do not:

“I don’t know him all that well, so I don’t want to vouch for him too much.”

“If she was in this room, I wouldn’t have a clue who she was.”

“I recognise my [negative] input is based on nothing as I don’t know her.”

“I think I’m supposed to be her mentor …”

This is bound up in all sorts of biases: the fact that majority employees have significantly more mentors and sponsors than minority employees; that attribution errors kick in and it becomes all too easy to assume that if someone hasn’t got an advocate in the room then it is somehow their fault; the fact that we all, including these decision makers, find it easier to build stronger, trusting relationships with people that are like ourselves.

Simply because people know that this is an issue does not mean they have become savvy enough to guard against its impact.

Overcoming our biases

These are just four of the many biases and conditions for bias that we see consistently in these discussions. And two main groups of employees are consistently affected: women and introverts.

That’s not to say that other minority groups aren’t affected – it varies very much organisation to organisation, with some organisations having clear levels of bias and mistrust towards employees from BAME groups and other nationalities, for example.

So what can we do about it? Simple steps do make a difference – clarifying the criteria so that there are fewer borderline employees to start with will help. Focusing more time on the borderline employees / candidates, and less time on those where the whole room is in ready agreement about would also help. As would simple rules – either everyone in the room has an advocate or, if not, we don’t allow personal advocacy statements.

Fundamentally, awareness is a critical step, but it needs to be more than a one-off workshop on bias that people may or may not attend.

Conducting live bias reviews is key, as it holds up a mirror and helps people see how biases play out in reality in the very meetings they are part of.

That way we can make significantly bigger strides in overcoming another common bias – blind spot bias – where people believe that they are not part of the problem, as illustrated by the investment banker.

Want to learn more about unconscious bias?

Read Richard Chapman-Harris’ article on whether cultural fit is just another form of unconscious bias here.