Monday, October 29, 2012

Learning to Make Bad Choices

Every day, people make choices among options they may have never experienced. Yet, they can make such choices rapidly, even confidently—yet unwisely. Whether good or bad, the choices are likely to be biased from past learning conditions,

Research is now clarifying how biased choices are made. All choices, biased or not, are influenced by the values we place on alternative choice options. If a past choice led to a good outcome, people are biased to make that choice again because a positive value has been attached to it.

If in a current situation where there is no past relevant experience, we can still make a choice easily. To do that, we must be able to assign positive value to the choice made. But how can that be done in the absence of past experience? It now appears from recent experiments that such decisions can guided by past memory associations.

Memories certainly modulate value assignment and thus decision-making. In one recent study, brain-scan imaging of humans has revealed that giving people monetary rewards activates a pre-established network of memories that spread the positive value of reward to non-rewarded items stored in memory. This creates a bias for later decisions to choose non-rewarded items.

As an aside, the study revealed that a key structure involved in memory formation, the hippocampus, predicts choice bias. It appears that activity in the hippocampus, as revealed in brain scans, helps to spread the reward value among memories of items that had never been rewarded, and thus bias future choices for these non-rewarded items.

The rationale for the study began with the authors’ realization of how the hippocampus helps to form memories. It encodes relationships between items and events to form a remembered association. Later, when an item is recalled, the hippocampus can automatically reactive the neural representation of the associated item. Here then might be a way that the hippocampus could re-allocate reward value when a past situation is recalled.

The choice task had three stages. First, the subjects had to learn association between paired items. They were shown a series of pairs of images, first one and then the other. Next was a reward phase, in which half of the second stimuli were rewarded with money. Thus, subjects  unconsciously learned a response to value those stimuli that were directly rewarded.  The researchers expected that the non-rewarded member of the pair might pick up some value by virtue of being associated with the rewarded member of the pair. If so, a bias might have been created for these non-rewarded items in future choice situations. Indeed, that is what they found in a later test in which only the first member of each pair was presented and subjects were asked to pick the preferred items. If no transfer of value had occurred in the prior learning, the choices would have been evenly divided. But that is not what was found. Subjects more often than chance preferred the stimuli that had previously been paired with a rewarded second stimulus.

Why is this important to the issue of blame? This issue was never considered by the authors. But to extrapolate from the experiment, we could conclude that sometimes we may blame a bad choice on the wrong cause. The real cause may have been our own bias for making a bad choice, in which we had erroneously assigned value on the basis of some past associations with experiences that were rewarded. That bias may have been created by some past experiences in which we had inappropriately assigned a high value to certain choice options that did not deserve such value.

So, for example, we might have had a past experience in which a good thing happened at the same time as other events which had no learned value or even negative value. But the value of the good thing spreads to bias attitudes toward the other events. In the future, in a different circumstance, we might unwisely choose the previously low-value option.

Here is a common example. Suppose we have a dear friend who we have found to offer a valuable relationship. We meet the dear friend’s associates, who may be unseemly characters that would make poor choices for us to value as friends. Yet, the value we attached to the dear friend spills over to the others and makes us less likely to be critical of them. Thus, we might start running with a bad crowd of people that get us into trouble. When we get into trouble, we likely will place the blame in the wrong place, without realizing it all started with the bias created in us by the value we had placed in our dear friend.

You can probably conjure your own examples. This whole process of learned bias can affect every aspect of our life. It is important for us to be more aware of our biases and the true value of our choice options.


Wimmer, G. E. and Shohamy, D. (2012). Preference by association: how memory mechanisms in the hippocampus bias decisions. Science. 338: 270-273.

No comments:

Post a Comment