For 95% of the history of modern humans we were exclusively hunter gatherers. Then suddenly about 12,000 years ago, something happened that revolutionised the way humans lived and enabled the complex societies we have today: farming.
But what triggered this revolution? Understanding this is incredibly challenging – because this occurred so far in the past, there are many factors to consider. However, by simulating the past using a complex computational model, we found that the switch from foraging to farming most likely began with very small groups of people that were using the concept of property rights.
Farming: an unlikely choice
It may seem obvious why we switched from foraging to farming: it made it possible to stay in one place, feed larger populations, have greater food security and build increasingly complex societies, political structures, economies and technologies. However, these advantages took time to develop and our early farmer ancestors would not have seen these coming.
Indeed, archaeological research suggests that when farming began it was not a particularly attractive lifestyle. It involved more work, a decrease in the quality of nutrition and health, an increase in disease and infection, and greater challenges in defending resources. For a hunter-gatherer at the cusp of the “agricultural revolution”, a switch to farming wasn’t the obvious choice.
So why then did early farmers choose this lifestyle? We know that farming started around the beginning of the Holocene, the unusually warm and stable climate phase we find ourselves in today. This may have lead to changes in population sizes, mobility and resource quality and predictability – and in turn changes to social organisation. We decided to examine the issue using a new method to analyze a computational model on the origins of our farming past.
A couple of years ago, a well-regarded model of the origins of agriculture considered a population of people divided into groups who first gained food by foraging or farming, and then either tried to share or steal each other’s food. Whether they decided to share or steal was determined by their behavioural strategy: sharer, civic or bourgeois. Sharers always share, of course; civics are sharer-enforcers – where they will punish someone who tries to not share; and if the agent is a bourgeois then they will never try to steal farmed food, so they could be considered as having farming-friendly property rights.
According to the model, the agents occasionally changed their behaviour and also sometimes migrated to a different group. For each cycle the amount of food gained from farming changed according to climate volatility, which was estimated using ice core data given in intervals of 20 years.
These simulations showed that farming and farming-friendly property rights evolved together around the same time as farming is observed in the archaeological record. However, it included many factors (parameters) other than property rights that may have influenced the emergence of farming.
Millions of possibilities
In our study we explored the effects of other parameters in more detail. Knowing values for parameters that shaped something that occurred so long ago is very difficult. For example, how can you know what proportion of ancient people migrated from group to group in each generation? You may find in ethnographic or archaeological data (if you are very lucky) that this is approximately a fifth of the population at a certain time. But what if it was a sixth? Would this lead to a different outcome?
And what about the simultaneous effects of another parameter, for example the amount of food gained by foraging, for which you don’t know the exact value either. You would need to check all the different combinations of reasonable values for these two parameters on the outcome of the model. This is complicated enough, but at least you can visualise the results. Now try adding a third unknown parameter – or ten. Things can get very complicated very quickly.
We implemented a method that explores the effects of 11 parameters: the number of groups; the migration rate; the amount of behavioural experimentation; the cost of a conflict within the group; the cost and probability of a conflict between groups; the foraging productivity; the intensities of farming and foraging property rights; the amount of farming investment; and the level of conformism. This method randomly picked 12m different combinations of these parameter values from realistic ranges. We then ran the model for each of these combinations and asked how “good” the result found was; this was defined as how many of the people in the model were farmers by 9,000 years ago (when farming became established).
We then looked back at the combinations of parameters that made these “good” results. From our results we could predict that for farming to develop it would have been beneficial for groups to be small, around four to six people, and that property rights were indeed an important factor. We also found that it is possible for farming to have begun even if the amount of food gained by farming was less than that of foraging.
Statistician George Box once said that “all models are wrong”. But when we try to understand processes that are shaped by many factors our intuitions will usually fail us. All models maybe wrong, but they can guide and sharpen those intuitions.