Does a green environment lead to less impulsive decisions?

If you were offered the choice between 100 euros straight away or 110 euros in three months from now, what would you choose? Given low interest rates, it seems more rational to go for the larger amount in due time, but many people will take the directly available money. Professor Mark van Vugt and his team of the Vrije Universiteit of Amsterdam tried to find out whether it would make a difference in what kind of environment this question was posed to people. Prof. Van Vugt : “Impulsivity and short term thinking are typical human features. In our study, we investigated whether nature could affect those mechanisms in our brain.’ (VU press release, Nov. 6^th 2013).

The study is published as: Do natural landscapes reduce future discounting in humans? Arianne J. van der Wal, Hannah M. Schade, Lydia Krabbendam, and Mark van Vugt (Proc R Soc B 2013 280: 20132295).

So why would their findings matter to us? The researchers answer this question:

Many of the problems that society faces today have to do with the typical human tendency to value the present more than the future. For instance, will I eat my cake now or exercise first and then eat it? Do I want to start a family now or wait until I have finished my education? Do I fill up my car with petrol or purchase an electrical car?

Our results show that giving people the opportunity to immerse themselves in a green environment may shift their time horizon from the here and now to the future. With the majority of the citizens of this planet now living in large, urban areas we must find clever ways to bring people in regular contact with nature. Particularly for our children regular outdoor experiences may be important to as they are more likely to discount the future.

That’s quite a big claim and I’m not that sure this research shows any of this. It looks kinda flimsy to me. Two of the three experiments were similar to the priming experiments in social psychology, which have come under a lot of criticism the last couple of years (especially after the Stapel fraude case). My main problem with these kind of experiments is that the researchers assume that they prime the participants in a way they control. But there are generally many things you can think of what is actually happening and it doesn’t have to be the same for all participants.
Therefore the third experiment they did, seems more interesting. It didn’t involve pictures but gave people the actual feel of the environment. The researchers let them walk around in a wood close to Amsterdam or in a urban area (business district Amsterdam Zuidas). After a five minute walk the participants met with a research assistant who put some tasks to them. The task which gets the most attention in the media is like a game: first they were asked what they would prefer, 100 euros now or 110 euros in 90 days from now. The participants who took the 100 euros now, were asked a follow-up question, 100 euros now or 120 euros after 90 days. And so on. This would end at 100 euros now or 170 euros later.The idea is that people who choose the delayed but higher amount are also more concerned about long term issues in general. It’s called a temporal discounting game and has been used a lot by other researchers. But here, it seems to me, the execution of the game was somewhat different and the authors didn’t give that enough thought.

The hypothesis was that the group in the natural environment would be more easily convinced to go for the delayed reward, because … Well that’s the real question of course, but the researchers assume (and I think it seems plausible to a lot of people) that exposure to nature does make you think more about future consequences and therefore lead to a more thoughtful decision. In the wording of the article: ‘when people are exposed to scenes of natural landscapes their discount rates are lower compared with exposure to urban landscapes.’

The results are given as follows:

Confirming Hypothesis 1, participants in the nature versus urban condition showed a significant difference in temporal discounting (F_1,41 = 5.41, p = 0.025, part. η² = 0.12). In the nature condition, the individual indifference point of the participants was lower than that in the urban condition, with M_nature = 122.38 (s.d. = 16.40) and M_urban = 135.45 (s.d. = 20.17). On average, we found a 10% reduction in future discounting in the nature condition versus urban condition (figure 3).

and presented in a graph:

To me this presentation seems unnecessarily complicated. Why not a simple table which shows how many of the participants eventually ran away with which amount of money? Luckily the data of this experiment are also given (hurray for Open Access!) and that enabled me to make such a table myself:

Something odd turns up: it seems that there were two (urban) participants who managed to get away with 180 euros, while we were told that the last bid was ‘100 now or 170 after three months’. I guess that these two were rather clever participants who figured out that they were getting offered better bids every turn and that it would make a lot of sense to sit out for a while. Unfortunately the last bid was 170 euros and they didn’t take it. [also see the update below this blog]
The researchers tell us that this ‘game’ was used by other researchers (Wilson & Daily, Kirby et al.), but there’s a important difference. Here the low amount is 100 euros every turn, while in the other experiments also the lower amount differed every turn. So after a first offer of ‘100 now or 110 later’ it could be followed by something like ’85 now or 102 later’. Then the participants could not assume that waiting probably increases their maximum return without risk of falling below 100 euros (which is a nice return anyway and why not try to get really lucky if your day has already turned out fine?).

In the averages the contribution of these two is taken as 180 euros, but that makes no sense (is it a coding error?). If you look at it as a proper game, their score should have been 100 euros. And otherwise, we don’t know at what point they would have grabbed the higher delayed amount. Stopping the offers at 170 euros is a rather random choice. The averages don’t have a real meaning and I think you can’t draw any conclusions from those. Calculating standard deviations and using a statistical test ( which seems inappropriate anyway) to compare the two groups therefore also seems a bit silly to me.
However, the data still point to a difference if you look at it with more care. The ‘game’ aspect can probably be ignored if you only look at the first question (‘100 euros now or 110 after 3 months’). It is still a significant difference then: 4 out of 22 urban participants versus 11 out of 21 participants in the woods (Fisher’s exact test: p=0.02).

But there are also other issues. Science journalist Elmar Veerman, via whom I became aware of this research, wrote a nice article about this (in Dutch). He raises the question whether this test could actually be more about measuring the level of trust the participants have in the person offering the money than about discounting the future. It is also strange that no difference was found between participants who grew up in natural or urban environments. That it is just a short term effect seems a bit unlikely if the effect really exists in the way the researchers think it does.
Another point, acknowledged by prof. van Vugt, is that the nature condition was ‘nice’. It would be interesting to see whether the same results would show when more harsh natural conditons were used (a desert, an area with dangerous animal living in it). In short, was the ‘nature’ condition’ really about nature as such? What if they would have compared with a group walking in a more pitoresque cosy urban environment like there are plenty in Amsterdam as well?

Of course the article is about more than just this experiment and I only looked at the most interesting experiment from my viewpoint [update 15/11/2013: just noticed that there are similar problems with the data of the first experiment]. So I won’t damn this article as a whole, but there are certainly some issues the authors should look at, I think. Also I don’t have a big problem with this kind of research and faults can be made everywhere. But what does worry me are the press releases of universities which draw attention to it. With the language used and the way they translate the scientific results, they claim far more than what the scientific articles show, but we see those interpretations parotted in the media all over.

 

Update 17-1-2014 First author of the study, Arianne van der Wal, has published an eLetter on the article (Jan 7th 2014). From this: Within 2 cases in the Fieldstudy and 4 cases in Experiment 1, participants did not even prefer the maximum of 170 euro in 90 days over 100 euro now. This means that their indifference point lies at least with 180 euro in 90 days (which would have been the following question in line within the temporal discounting game). That is why the indifference point is determined as 180 for these participants. It doesn’t address the issues I have raised in my blog: this ‘180 euros’ is different than the other outcomes of the ‘game’. The 180’s are not depicted in the graphs but they are used in calculating the means.