Avoiding costly mistakes in groups: The evolution of error management in collective decision making

Individuals continuously have to balance the error costs of alternative decisions. A wealth of research has studied how single individuals navigate this, showing that individuals develop response biases to avoid the more costly error. We, however, know little about the dynamics in groups facing asymmetrical error costs and when social influence amplifies either safe or risky behavior. Here, we investigate this by modeling the decision process and information flow with a drift–diffusion model extended to the social domain. In the model individuals first gather independent personal information; they then enter a social phase in which they can either decide early based on personal information, or wait for additional social information. We combined the model with an evolutionary algorithm to derive adaptive behavior. We find that under asymmetric costs, individuals in large cooperative groups do not develop response biases because such biases amplify at the collective level, triggering false information cascades. Selfish individuals, however, undermine the group’s performance for their own benefit by developing higher response biases and waiting for more information. Our results have implications for our understanding of the social dynamics in groups facing asymmetrical errors costs, such as animal groups evading predation or police officers holding a suspect at gunpoint.     

Maternal Oxytocin Is Linked to Close Mother-Infant Proximity in Grey Seals (Halichoerus grypus)

Maternal behaviour is a crucial component of reproduction in all mammals; however the quality of care that mothers give to infants can vary greatly. It is vital to document variation in maternal behaviour caused by the physiological processes controlling its expression. This underlying physiology should be conserved throughout reproductive events and should be replicated across all individuals of a species; therefore, any correlates to maternal care quality may be present across many individuals or contexts. Oxytocin modulates the initiation and expression of maternal behaviour in mammals; therefore we tested whether maternal plasma oxytocin concentrations correlated to key maternal behaviours in wild grey seals (Halichoerus grypus). Plasma oxytocin concentrations in non-breeding individuals (4.3 ±0.5 pg/ml) were significantly lower than those in mothers with dependent pups in both early (8.2 ±0.8 pg/ml) and late (6.9 ±0.7 pg/ml) lactation. Maternal plasma oxytocin concentrations were not correlated to the amount of nursing prior to sampling, or a mother’s nursing intensity throughout the dependant period. Mothers with high plasma oxytocin concentrations stayed closer to their pups, reducing the likelihood of mother-pup separation during lactation which is credited with causing starvation, the largest cause of pup mortality in grey seals. This is the first study to link endogenous oxytocin concentrations in wild mammalian mothers with any type of maternal behaviour. Oxytocin’s structure and function is widely conserved across mammalian mothers, including humans. Defining the impact the oxytocin system has on maternal behaviour highlights relationships that may occur across many individuals or species, and such behaviours heavily influence infant development and an individual’s lifetime reproductive success.     

The legacy effects of keystone individuals on collective behaviour scale to how long they remain within a group

The collective behaviour of social groups is often strongly influenced by one or few individuals, termed here ‘keystone individuals’. We examined whether the influence of keystone individuals on collective behaviour lingers after their departure and whether these lingering effects scale with their tenure in the group. In the social spider, Stegodyphus dumicola, colonies'' boldest individuals wield a disproportionately large influence over colony behaviour. We experimentally manipulated keystones'' tenure in laboratory-housed colonies and tracked their legacy effects on collective prey capture following their removal. We found that bolder keystones caused more aggressive collective foraging behaviour and catalysed greater inter-individual variation in boldness within their colonies. The longer keystones remained in a colony, the longer both of these effects lingered after their departure. Our data demonstrate that, long after their disappearance, keystones have large and lasting effects on social dynamics at both the individual and colony levels.     

Opinion Formation by Social Influence: From Experiments to Modeling

Predicting different forms of collective behavior in human populations, as the outcome of individual attitudes and their mutual influence, is a question of major interest in social sciences. In particular, processes of opinion formation have been theoretically modeled on the basis of a formal similarity with the dynamics of certain physical systems, giving rise to an extensive collection of mathematical models amenable to numerical simulation or even to exact solution. Empirical ground for these models is however largely missing, which confine them to the level of mere metaphors of the real phenomena they aim at explaining. In this paper we present results of an experiment which quantifies the change in the opinions given by a subject on a set of specific matters under the influence of others. The setup is a variant of a recently proposed experiment, where the subject’s confidence on his or her opinion was evaluated as well. In our realization, which records the quantitative answers of 85 subjects to 20 questions before and after an influence event, the focus is put on characterizing the change in answers and confidence induced by such influence. Similarities and differences with the previous version of the experiment are highlighted. We find that confidence changes are to a large extent independent of any other recorded quantity, while opinion changes are strongly modulated by the original confidence. On the other hand, opinion changes are not influenced by the initial difference with the reference opinion. The typical time scales on which opinion varies are moreover substantially longer than those of confidence change. Experimental results are then used to estimate parameters for a dynamical agent-based model of opinion formation in a large population. In the context of the model, we study the convergence to full consensus and the effect of opinion leaders on the collective distribution of opinions.     

Food sharing is linked to urinary oxytocin levels and bonding in related and unrelated wild chimpanzees

Humans excel in cooperative exchanges between unrelated individuals. Although this trait is fundamental to the success of our species, its evolution and mechanisms are poorly understood. Other social mammals also build long-term cooperative relationships between non-kin, and recent evidence shows that oxytocin, a hormone involved in parent–offspring bonding, is likely to facilitate non-kin as well as kin bonds. In a population of wild chimpanzees, we measured urinary oxytocin levels following a rare cooperative event—food sharing. Subjects showed higher urinary oxytocin levels after single food-sharing events compared with other types of social feeding, irrespective of previous social bond levels. Also, urinary oxytocin levels following food sharing were higher than following grooming, another cooperative behaviour. Therefore, food sharing in chimpanzees may play a key role in social bonding under the influence of oxytocin. We propose that food-sharing events co-opt neurobiological mechanisms evolved to support mother–infant bonding during lactation bouts, and may act as facilitators of bonding and cooperation between unrelated individuals via the oxytocinergic system across social mammals.     

Robots with Display Screens: A Robot with a More Humanlike Face Display Is Perceived To Have More Mind and a Better Personality

It is important for robot designers to know how to make robots that interact effectively with humans. One key dimension is robot appearance and in particular how humanlike the robot should be. Uncanny Valley theory suggests that robots look uncanny when their appearance approaches, but is not absolutely, human. An underlying mechanism may be that appearance affects users’ perceptions of the robot’s personality and mind. This study aimed to investigate how robot facial appearance affected perceptions of the robot’s mind, personality and eeriness. A repeated measures experiment was conducted. 30 participants (14 females and 16 males, mean age 22.5 years) interacted with a Peoplebot healthcare robot under three conditions in a randomized order: the robot had either a humanlike face, silver face, or no-face on its display screen. Each time, the robot assisted the participant to take his/her blood pressure. Participants rated the robot’s mind, personality, and eeriness in each condition. The robot with the humanlike face display was most preferred, rated as having most mind, being most humanlike, alive, sociable and amiable. The robot with the silver face display was least preferred, rated most eerie, moderate in mind, humanlikeness and amiability. The robot with the no-face display was rated least sociable and amiable. There was no difference in blood pressure readings between the robots with different face displays. Higher ratings of eeriness were related to impressions of the robot with the humanlike face display being less amiable, less sociable and less trustworthy. These results suggest that the more humanlike a healthcare robot’s face display is, the more people attribute mind and positive personality characteristics to it. Eeriness was related to negative impressions of the robot’s personality. Designers should be aware that the face on a robot’s display screen can affect both the perceived mind and personality of the robot.     

The Undecided Have the Key: Interaction-Driven Opinion Dynamics in a Three State Model

The effects of interpersonal interactions on individual’s agreements result in a social aggregation process which is reflected in the formation of collective states, as for instance, groups of individuals with a similar opinion about a given issue. This field, which has been a longstanding concern of sociologists and psychologists, has been extended into an area of experimental social psychology, and even has attracted the attention of physicists and mathematicians. In this article, we present a novel model of opinion formation in which agents may either have a strict preference for a choice, or be undecided. The opinion shift emerges, in a threshold process, as a consequence of a cumulative persuasion for either one of the two opinions in repeated interactions. There are two main ingredients which play key roles in determining the steady states: the initial fraction of undecided agents and the change in agents’ persuasion after each interaction. As a function of these two parameters, the model presents a wide range of solutions, among which there are consensus of each opinion and bi-polarization. We found that a minimum fraction of undecided agents is not crucial for reaching consensus only, but also to determine a dominant opinion in a polarized situation. In order to gain a deeper comprehension of the dynamics, we also present the theoretical framework of the model. The master equations are of special interest for their nontrivial properties and difficulties in being solved analytically.     

The financial cost of status signaling: Expansive postural displays are associated with a reduction in the receipt of altruistic donations

Models of human altruism suggest that decisions to help are influenced by assessments of both potential recipients' need state and their competence, as high need increases the value of gifts received, and competent recipients can most effectively use and repay gifts. Need and competence are often inversely related, however, raising the question of how altruists weigh these competing sources of information. We examined the impact of a nonverbal display (expansive posture) that, by signaling high status, simultaneously cues both low need and high competence, on actual altruistic behaviors: donations of financial aid to needy individuals. Across three studies using ecologically valid data drawn from a micro-lending charity website, men who displayed expansive posture while requesting aid faced a substantial reduction in the amount of aid they received; this effect held controlling for a range of relevant covariates. These findings demonstrate that: (a) altruists bias their giving toward those in greater need rather those who may be more competent, and (b) subtle nonverbal cues of status influence altruistic decision-making.     

Reverse Correlating Love: Highly Passionate Women Idealize Their Partner’s Facial Appearance

A defining feature of passionate love is idealization—evaluating romantic partners in an overly favorable light. Although passionate love can be expected to color how favorably individuals represent their partner in their mind, little is known about how passionate love is linked with visual representations of the partner. Using reverse correlation techniques for the first time to study partner representations, the present study investigated whether women who are passionately in love represent their partner’s facial appearance more favorably than individuals who are less passionately in love. In a within-participants design, heterosexual women completed two forced-choice classification tasks, one for their romantic partner and one for a male acquaintance, and a measure of passionate love. In each classification task, participants saw two faces superimposed with noise and selected the face that most resembled their partner (or an acquaintance). Classification images for each of high passion and low passion groups were calculated by averaging across noise patterns selected as resembling the partner or the acquaintance and superimposing the averaged noise on an average male face. A separate group of women evaluated the classification images on attractiveness, trustworthiness, and competence. Results showed that women who feel high (vs. low) passionate love toward their partner tend to represent his face as more attractive and trustworthy, even when controlling for familiarity effects using the acquaintance representation. Using an innovative method to study partner representations, these findings extend our understanding of cognitive processes in romantic relationships.     

Decision Rules and Group Rationality: Cognitive Gain or Standstill?

Recent research in group cognition points towards the existence of collective cognitive competencies that transcend individual group members’ cognitive competencies. Since rationality is a key cognitive competence for group decision making, and group cognition emerges from the coordination of individual cognition during social interactions, this study tests the extent to which collaborative and consultative decision rules impact the emergence of group rationality. Using a set of decision tasks adapted from the heuristics and biases literature, we evaluate rationality as the extent to which individual choices are aligned with a normative ideal. We further operationalize group rationality as cognitive synergy (the extent to which collective rationality exceeds average or best individual rationality in the group), and we test the effect of collaborative and consultative decision rules in a sample of 176 groups. Our results show that the collaborative decision rule has superior synergic effects as compared to the consultative decision rule. The ninety one groups working in a collaborative fashion made more rational choices (above and beyond the average rationality of their members) than the eighty five groups working in a consultative fashion. Moreover, the groups using a collaborative decision rule were closer to the rationality of their best member than groups using consultative decision rules. Nevertheless, on average groups did not outperformed their best member. Therefore, our results reveal how decision rules prescribing interpersonal interactions impact on the emergence of collective cognitive competencies. They also open potential venues for further research on the emergence of collective rationality in human decision-making groups.     

The volunteer's dilemma and the optimal size of a social group

If one or few individuals are enough to perform an action that produces a collective good and if this action has a cost, living in group can be beneficial because the cost can be shared with other individuals. Without coordination, however, the production of a collective good by the contribution of one or few individuals is inefficient and can be modelled as a volunteer's dilemma. In the volunteer's dilemma the individuals that pay the cost for the production of the collective good benefit from their action if nobody else volunteers, but the cost is wasted if too many individuals volunteer. Increasing group size reduces the need of volunteering for each member of the group; the overall benefit for the group, however, decreases too because the larger the group is, the less likely it is that the collective good is produced. This problem persists even with a high degree of relatedness between group members; an optimal, intermediate group size exists that maximizes the probability to produce the collective good.     

Horizontal DNA Transfer Mechanisms of Bacteria as Weapons of Intragenomic Conflict

Horizontal DNA transfer (HDT) is a pervasive mechanism of diversification in many microbial species, but its primary evolutionary role remains controversial. Much recent research has emphasised the adaptive benefit of acquiring novel DNA, but here we argue instead that intragenomic conflict provides a coherent framework for understanding the evolutionary origins of HDT. To test this hypothesis, we developed a mathematical model of a clonally descended bacterial population undergoing HDT through transmission of mobile genetic elements (MGEs) and genetic transformation. Including the known bias of transformation toward the acquisition of shorter alleles into the model suggested it could be an effective means of counteracting the spread of MGEs. Both constitutive and transient competence for transformation were found to provide an effective defence against parasitic MGEs; transient competence could also be effective at permitting the selective spread of MGEs conferring a benefit on their host bacterium. The coordination of transient competence with cell–cell killing, observed in multiple species, was found to result in synergistic blocking of MGE transmission through releasing genomic DNA for homologous recombination while simultaneously reducing horizontal MGE spread by lowering the local cell density. To evaluate the feasibility of the functions suggested by the modelling analysis, we analysed genomic data from longitudinal sampling of individuals carrying Streptococcus pneumoniae. This revealed the frequent within-host coexistence of clonally descended cells that differed in their MGE infection status, a necessary condition for the proposed mechanism to operate. Additionally, we found multiple examples of MGEs inhibiting transformation through integrative disruption of genes encoding the competence machinery across many species, providing evidence of an ongoing “arms race.” Reduced rates of transformation have also been observed in cells infected by MGEs that reduce the concentration of extracellular DNA through secretion of DNases. Simulations predicted that either mechanism of limiting transformation would benefit individual MGEs, but also that this tactic’s effectiveness was limited by competition with other MGEs coinfecting the same cell. A further observed behaviour we hypothesised to reduce elimination by transformation was MGE activation when cells become competent. Our model predicted that this response was effective at counteracting transformation independently of competing MGEs. Therefore, this framework is able to explain both common properties of MGEs, and the seemingly paradoxical bacterial behaviours of transformation and cell–cell killing within clonally related populations, as the consequences of intragenomic conflict between self-replicating chromosomes and parasitic MGEs. The antagonistic nature of the different mechanisms of HDT over short timescales means their contribution to bacterial evolution is likely to be substantially greater than previously appreciated.     

Are You Your Friends’ Friend? Poor Perception of Friendship Ties Limits the Ability to Promote Behavioral Change

Persuasion is at the core of norm creation, emergence of collective action, and solutions to ‘tragedy of the commons’ problems. In this paper, we show that the directionality of friendship ties affect the extent to which individuals can influence the behavior of each other. Moreover, we find that people are typically poor at perceiving the directionality of their friendship ties and that this can significantly limit their ability to engage in cooperative arrangements. This could lead to failures in establishing compatible norms, acting together, finding compromise solutions, and persuading others to act. We then suggest strategies to overcome this limitation by using two topological characteristics of the perceived friendship network. The findings of this paper have significant consequences for designing interventions that seek to harness social influence for collective action.     

Analysing the Unequal Effects of Positive and Negative Information on the Behaviour of Users of a Taiwanese On-Line Bulletin Board

The impact of social influence causes people to adopt the behaviour of others when interacting with other individuals. The effects of social influence can be direct or indirect. Direct social influence is the result of an individual directly influencing the opinion of another, while indirect social influence is a process taking place when an individual’s opinion and behaviour is affected by the availability of information about others’ actions. Such indirect effect may exhibit a more significant impact in the on-line community because the internet records not only positive but also negative information, for example on-line written text comments. This study focuses on indirect social influence and examines the effect of preceding information on subsequent users’ opinions by fitting statistical models to data collected from an on-line bulletin board. Specifically, the different impacts of information on approval and disapproval comments on subsequent opinions were investigated. Although in an anonymous situation where social influence is assumed to be at minimum, our results demonstrate the tendency of on-line users to adopt both positive and negative information to conform to the neighbouring trend when expressing opinions. Moreover, our results suggest unequal effects of the local approval and disapproval comments in affecting the likelihood of expressing opinions. The impact of neighbouring disapproval densities was stronger than that of neighbouring approval densities on inducing subsequent disapproval relative to approval comments. However, our results suggest no effects of global social influence on subsequent opinion expression.     

Money,power, and mitochondria

Can we devise ways for the ultra‐rich to put their wealth to good use?

The world’s richest individuals and their wealth have recently come under the spotlight, with all those luxury yachts apparently owned by Russian oligarchs and their political friends veering to safe havens beyond the reach of sanctions; most of them vastly surpassing in value the cost of an advanced warship or the on‐off‐on saga of Elon Musk’s purchase of Twitter for a sum that exceeds the annual GDP of a mid‐sized country.Even if their riches have been acquired through legal means and provide joy and sustenance to many, how can it be justified that a small number of individuals own such vast assets, when a substantial fraction of the world’s population still lives in abject misery without even the benefit of adequate food and clean water?At this time, molecular biologists have yet to join the ranks of the ultra‐rich. But there is no reason to suppose that being a molecular biologist somehow insulates us from the greed and megalomania that seems to infect those who become billionaires on the back of their inventions and investments. The most likely playgrounds in which a hypothetical molecular biologist might acquire such riches are in biotech and via the pharmaceutical industry. Big Pharma already enjoys low public esteem—in my view somewhat unfairly because its critics rarely take into account the real cost of R&D required to develop new drugs. But it is an easy target nonetheless, not least because of its scale, which inflates the bottom line of balance sheets in ways not easily justified in the eyes of “Joe Public”. Not to mention the many drugs that allegedly do not really work. In the public mind, we are already allied with corporate greed. It is not hard to see how this could become a reality.Those who have accumulated extreme wealth in other sectors, and accrued power and influence to go with it, are viewed unfavorably by most people, even if some super‐rich individuals are admired for their creativity (Walker et al, 2021). Obscene wealth is still considered obscene, even if some of it is invested in scientific research, in the alleviation of human suffering, or in sponsorship of the arts. The fact that the ultra‐rich are able to do so seems to many like a usurpation of what is rightfully the responsibility of government, especially in democratic countries, and what do the wealthiest actually spend their money on anyway? Even if some of it is given away to charity, it is at the arbitrary whim of one or a few individuals, and not a transparent process with public accountability. Surely, it would be far better if they were all subject to a global wealth tax, the proceeds of which might support the UN, whose current annual budget (about US$ 3 billion) is mere pocket money for the world’s wealthiest persons.But how could such a scheme actually work? Although the widespread sanctioning of those who supposedly bankroll the Kremlin seems to have had some effect, the ultra‐rich invariably devise ways of circumventing legal obstacles thrown in their path. There is a widespread suspicion that economic sanctions ultimately punish ordinary people rather than wayward decision‐makers or their wealthy friends. High taxes promote tax avoidance and tax havens. Strict regulations to prevent money laundering just fuel the rise of cryptocurrencies. Laws beget loopholes.Global legislation with global implementation may look like an attractive tool, but its track record is not impressive. Consider the Kyoto/Paris process on climate change, for example, or the proposal for a 15% minimum corporate tax rate worldwide. Its effects are likely to be blunted by the floor effectively becoming the ceiling, so that most or all corporations could end up paying less and not more in taxes.What about self‐regulation? Suppose EMBO made a rule that required any individual member to donate wealth accrued from molecular biology inventions above some threshold to charitable causes or be kicked out of the organization? Would such a person care? And if EMBO had any real power to frustrate their activities, would they not just be tempted to set up a rival organization where they would be free to continue their work without such restrictions?To limit the wealth and power of individuals made rich by the capitalist market economy in which we all live, I think there is only one viable tool at our disposal, namely, the capitalist market economy itself. If we collectively boycott the products and services of companies owned by the very rich, unless they donate the bulk of their wealth to good causes of which we approve, we could eventually shift the balance in favour of better practices.It may seem a pathetically inadequate tool and would certainly require a lot of time and effort. But in the long run, it could work. There is, for example, a growing movement to use market mechanisms to support ethical and environmentally sustainable businesses. Successful companies have adopted such advocacy in their own marketing, even if it sometimes seems a cynical ploy. They would not do so unless there is public demand. All we need is to add the accumulation of obscene wealth by company owners or majority shareholders as another abuse of the system that can potentially be reined in by consumer pressure. All it would take is for concerned individuals to devise a code of practice that could be used to judge compliance.Oh well, maybe in the end, I should just take comfort in the fact that all the power of the ultra‐rich stems ultimately from mitochondria.     

Duty or dream? Edwin G. Conklin's critique of eugenics and support for American individualism

This paper assesses ideas about moral andreproductive duty in American eugenics duringthe early twentieth century. While extremeeugenicists, including Charles Davenport andPaul Popenoe, argued that social leaders andbiologists must work to prevent individuals whowere ``unfit' from reproducing, moderates,especially Edwin G. Conklin, presented adifferent view. Although he was sympathetic toeugenic goals and participated in eugenicorganizations throughout his life, Conklinrealized that eugenic ideas rarely could meetstrict scientific standards of proof. Withthis in mind, he did not restrict his eugenicvision to hereditary measures. Relying onhis experience as an embryologist, Conklininstead attempted to balance more extremeeugenic claims – that emphasized the absolutelimits posed by heredity – with his own view of``the possibilities of development.' Throughhis critique he argued that most human beingsnever even begin to approach their hereditarypotential; he moderated his own eugenicrhetoric so that it preserved individualopportunity and responsibility, or what hasoften been labeled the American Dream.     

Midcell Recruitment of the DNA Uptake and Virulence Nuclease,EndA, for Pneumococcal Transformation

Genetic transformation, in which cells internalize exogenous DNA and integrate it into their chromosome, is widespread in the bacterial kingdom. It involves a specialized membrane-associated machinery for binding double-stranded (ds) DNA and uptake of single-stranded (ss) fragments. In the human pathogen Streptococcus pneumoniae, this machinery is specifically assembled at competence. The EndA nuclease, a constitutively expressed virulence factor, is recruited during competence to play the key role of converting dsDNA into ssDNA for uptake. Here we use fluorescence microscopy to show that EndA is uniformly distributed in the membrane of noncompetent cells and relocalizes at midcell during competence. This recruitment requires the dsDNA receptor ComEA. We also show that under ‘static’ binding conditions, i.e., in cells impaired for uptake, EndA and ComEA colocalize at midcell, together with fluorescent end-labelled dsDNA (Cy3-dsDNA). We conclude that midcell clustering of EndA reflects its recruitment to the DNA uptake machinery rather than its sequestration away from this machinery to protect transforming DNA from extensive degradation. In contrast, a fraction of ComEA molecules were located at cell poles post-competence, suggesting the pole as the site of degradation of the dsDNA receptor. In uptake-proficient cells, we used Cy3-dsDNA molecules enabling expression of a GFP fusion upon chromosomal integration to identify transformed cells as GFP producers 60–70 min after initial contact between DNA and competent cells. Recording of images since initial cell-DNA contact allowed us to look back to the uptake period for these transformed cells. Cy3-DNA foci were thus detected at the cell surface 10–11 min post-initial contact, all exclusively found at midcell, strongly suggesting that active uptake of transforming DNA takes place at this position in pneumococci. We discuss how midcell uptake could influence homology search, and the likelihood that midcell uptake is characteristic of cocci and/or the growth phase-dependency of competence.     

Oxytocin modulates selection of allies in intergroup conflict

In intergroup competition and conflict, humans benefit from coalitions with strong partners who help them to protect their in-group and prevail over competing out-groups. Here, we link oxytocin, a neuropeptide produced in the hypothalamus, to ally selection in intergroup competition. In a double-blind placebo-controlled experiment, males self-administered oxytocin or placebo, and made selection decisions about six high-threat and six low-threat targets as potential allies in intergroup competition. Males given oxytocin rather than placebo viewed high-threat targets as more useful allies and more frequently selected them into their team than low-threat targets.     

An initial ‘snapshot’ of sensory information biases the likelihood and speed of subsequent changes of mind

We often need to rapidly change our mind about perceptual decisions in order to account for new information and correct mistakes. One fundamental, unresolved question is whether information processed prior to a decision being made (‘pre-decisional information’) has any influence on the likelihood and speed with which that decision is reversed. We investigated this using a luminance discrimination task in which participants indicated which of two flickering greyscale squares was brightest. Following an initial decision, the stimuli briefly remained on screen, and participants could change their response. Using psychophysical reverse correlation, we examined how moment-to-moment fluctuations in stimulus luminance affected participants’ decisions. This revealed that the strength of even the very earliest (pre-decisional) evidence was associated with the likelihood and speed of later changes of mind. To account for this effect, we propose an extended diffusion model in which an initial ‘snapshot’ of sensory information biases ongoing evidence accumulation.