Methylmercury is a neurotoxic substance which is found at particularly high levels in predatory fish.  According to some studies, it is responsible for retarded neurological development in children.  As a precaution, in 2003, the Joint FAO/WHO Expert Committee on Food Additives reduced the recommended tolerable weekly intake of methylmercury for women of childbearing age, pregnant or breastfeeding women and young children.  At an international level, the circulation of information was chosen to manage this risk.  Recommendations differ from one country to another, notably as a function of national consumption.  The public health challenge consists in reducing the consumption of the most contaminated fish by the group at risk, without restricting the dietary intake of fish.  In France, the AFSSA published its recommendations in 2002 and the Ministry for Agriculture (DGAL) issued a press release in July 2006.
The experiment set up by the INRA researchers aimed to measure the effects of information on consumer behaviour by households, in the short and longer terms.  The households selected (201, representing 803 individuals) were followed for 5 months in 2005, and were asked to record their fish consumption (quantities and species) throughout the experiment.  At the end of the first month, they were randomized to two groups.  In the treatment group, the recommendations on consumption, aimed globally at restricting fish consumption to twice a week in the population at risk (see box below) was read by and left with all women in the households; the control group received no information.

Poor memorisation and the weak impact of information

At the end of the experiment, the mean weekly frequency of fish consumption (all species taken together) was higher than the recommended frequency of twice a week.  In total, 34% of women in both the treatment and control groups did not exceed two consumptions per week.
At the end of the experiment, few women succeeded in remembering the species referred to in the recommendations they had received.  Thus, although 50% of the women remembered the mention of fresh tuna, only 10% of them recalled that they should not consume it at all.

Strong anchorage of the positive values of fish

The recommendation given to the treatment group described the dangers of methylmercury and the nutritional benefits of fish.  A month after receiving this information, 50% of women considered that mercury constituted a major risk to their health; three months later, they were just 39%.
However, the positive values attributed to omega-3 oils did not decline and 70% of women at the end of the experiment continued to think that these fatty acids were beneficial to them and to other members of the household.  When the women were informed about the negative health effects of eating fish, they tended to underestimate the risk because of the strong anchorage of positive values, notably linked to the nutrition campaigns that have been widely disseminated in France for many years.
The results of this study led to a conclusion as to the weak effects of information on consumer behaviour.  Nevertheless, these results should be interpreted as arising from an experimental situation, with no account being taken of the social and family networks which are particularly important to consumer behaviour.  The weak impact of the recommendations raises several points concerning the information itself, particularly the  number of species cited and their associated frequencies, which were poorly memorised. 
At present, risk managers have drawn up different contents as a function of the countries targeted.  For example, the USA cites ten fish species that should be avoided, while in the French press release issued in 2006, only three species (swordfish, spearfish and siki) were mentioned, and readers were asked to refer to the Ministry of Agriculture website to obtain the full list of 28 wild predatory fish whose consumption should be restricted.  Some alternative solutions, such as the distribution of more succinct information (notably in terms of the species cited) and targeting pregnant women via the intermediary of the hospital system, could be considered.  Labelling on the packaging of the most contaminated species, specifying  for example "this product is not recommended for pregnant women and young children" could also be envisaged as a means of overcoming the poor memorisation of different species.

The results of this study show that the methods and conditions for circulating information need to be analysed more precisely.  In  addition, experimental methodologies may be useful with respect to improving the efficiency of government management concerning food risks.