In this paper we examine the impact of the Fukushima accident (March 2011) on public perceptions of nuclear power on a global scale. It is widely recognized that any future of nuclear power critically depends on public acceptance to sustain massive public subsidies. We will contrast conceptually and empirically two models of the 'Fukushima effect', an event & effect (EE) model (Kim, Kim & Kim, 2013) and our own challenge & response (CR) model. Firstly, we replicate Kim et al. (2013) who modelled retrospective opinion changes after March 2011 across 42 countries on a set of 'objective' predictors including geographical distance from Fukushima. But, instead of survey data ex-post-facto, we use historical opinion data 1996-2016 for 23+ countries. On historical data, the EE model has little explanatory power for opinion shifts, beyond the dependency on nuclear power in the energy mix. Secondly, we introduce the alternative CR model. Our hypothesis is that individual and societal responses to nuclear accidents are constrained by cultural memories. Memory, both individual and collective, is primarily adaptive and makes available schematic information to deal with novel situations. Memory creates familiarity and facilitates coping with uncertainty. The CR model introduces symbolic factors such as 'Past Responses to Nuclear Incidents', 'Nuclear Renaissance', and 'Long-term Acceptance Level' to explain the Fukushima effect of 2011.