Once a simulation of the world has been created, the brain uses it to predict what will happen next – how each object with a reference frame is expected to move. If new information from the sense organs matches with predictions from the simulation, then new signals from the sense organs are not cascaded far into the network but instead die out. We don’t experience them. In contrast, if the predictions of what will happen and observations diverge, then the simulation needs to be updated, often by drawing on information from neurons and cortical columns that had perhaps not recently contributed to building the simulation.
The way the brain compares predictions with observation is clever. For a neuron to fire, sending a signal down its axon and on to elsewhere in the network, it requires several synapses to fire, and for neighbouring neurons in the cortical column to be in the correct state. For a neuron to fire an electric signal down its axon, its soma needs to receive sufficient electrical signals from multiple synapses to push it over a threshold. If that threshold is not reached, the neuron enters a short period of time in a state where it is ready to fire if it receives another nudge. Neighbouring cells can detect this excited state, and this can prevent them from firing, a phenomenon known as lateral inhibition by neuroscientists. A consequence of this is that multiple neurons in a cortical column only simultaneously fire, sending signals throughout the brain, if they all experience many synapses firing simultaneously. This might happen if the simulation and new observation diverge – i.e. the simulation of the world is inaccurate.
Once a simulation of the world is running within our brains, the body may need to be directed to respond to events to remain out of harm’s way, or to acquire a resource. Some responses are entirely instinctive, such that an event in the simulation always results in a particular action. An example of such a response might be when a particular odour is sensed within the environment, and you always move away from it. Alternatively, we can use past experiences to decide on how to respond to a particular event in the simulation.
The area of the brain that brings together past experiences with the brain’s simulation of the world is the hippocampus. Like many scientific terms, the name comes from Greek. If you squint a little while looking at a diagram of the brain, the hippocampus takes the shape of a seahorse, which is the animal after which it is named. Neurons for memory can also run the simulations in our brain, and these are linked to networks that help you make a decision on how to behave. As with much of neuroscience, the details of how this happens are yet to be fully understood, but networks of memory and simulation neurons help us to make a decision. Once a decision is made, we act upon it, with signals sent to muscles about how we should move, or what we should say.
Consciousness researchers break down the steps from becoming aware of something in the outside world with our senses to responding to it as perception, attention, evaluation, integration, decision-making and action. From perceiving a ball thrown towards you, to moving your hands to catch it, happens quickly. Although the human brain is extraordinarily complicated, it is also remarkably efficient. But do our brains make good decisions? And how can we tell?
There is a whole field of scientific research about how to make decisions. Scientists working in this field research not only humans but also animals, and even artificial intelligence. You might be quite surprised at some of the insights that have been achieved. For example, consider the marriage problem, which asks what the best strategy is in choosing someone to marry. The problem has been attacked with computer simulations, but, as is often the case, some simplifying assumptions are needed to formulate the problem in such a way that a computer can help solve it. The marriage problem assumes that someone looking for a life partner dates prospective spouses randomly but sequentially, never two-timing a date. Everyone who is dated can be ranked, and there are no ties: for example, second place cannot be shared by two prospective partners. However, the person looking for a spouse does not know the ranking in advance. The problem also assumes that once a decision is made on whether to marry or move on, the decision is irreversible, such that the jilted person will be so hurt they will never agree to a second chance. Finally, assume that any marriage proposal will be accepted. Given these rules, when should you stop looking and settle for someone you like? Should you marry the first date you like, or should you keep looking just in case there is someone who would be a better life match? What is the best strategy to end up with the number-one-ranked partner?
The answer, it turns out, is a little involved. It depends upon the number of prospective dates, which of course in reality is something we probably do not know. The solution to the problem is also couched in terms of probability. It is not something like you should always choose the third date that you meet. Rather, it is based on the probability of choosing the number-one-ranked applicant if there are a particular number of dates and you have already rejected a certain number. For example, if there are six potential dates and you marry the third you meet, then there is a 42.8 per cent chance you chose the best option. There are stories, possibly apocryphal, of decision theorists using algorithms like this to select their life partners, but it is not how I decided to ask Sonya to marry me, and she assures me it is not how she reached the decision to say yes. We married because we make each other happy.
When decision-making theory is applied to animals, it is used to address questions such as, when should an individual move from one location to another? The animal is assumed to maximize something tangible, such as the amount of food it can eat in a certain location. When the returns of remaining in a patch are lower than the benefits of expending energy to move to another patch, the animal should move. The problem can be made more complicated by including social groups of animals, varying degrees of knowledge about likely food availability in other patches, and how different areas might vary in the risk of being predated. The application of decision-making theory to animal behaviour has provided interesting insights into how animals should behave, and sometimes they do seem to broadly follow the optimal strategy. However, working out what motivates an animal is not always straightforward. Not all animals spend all their time struggling to find food. They can be motivated by finding mates, water, shelter, warmth, safety, or even knowledge about their environment such as where competitors might be, and these motivations change with age, season and the weather. Despite the complexity, many decisions animals make appear to be optimal. They make many very good decisions.
Humans are animals, so it is not surprising that decision-making theory has also been widely applied to understand, and predict, how we behave. One difference between us and most animals is our ability to think about our long-term futures, and when we plan for these we try to make decisions that we think will make us most happy. Are our brains trying to maximize happiness? Happiness can be a bit of an elusive goal. Within a population, there are differences in how happy people are, and about a third of these differences can be attributed to genetics. Some people really are born happier than others. However, perhaps as much as two thirds of these differences are due to our circumstances and how we live our lives. Being financially secure helps with happiness. People living in poverty tend to be less happy than those who do not, but being super-rich doesn’t guarantee happiness either. Surveys of happiness across people on rich lists that ask participants questions about their state of well-being reveal they are only slightly more likely to be content than those who are not on the lists.