Age-related changes in midbrain dopaminergic regulation of the human reward system

The dopamine system, which plays a crucial role in reward processing, is particularly vulnerable to aging. Significant losses over a normal lifespan have been reported for dopamine receptors and transporters, but very little is known about the neurofunctional consequences of this age-related dopaminergic decline. In animals, a substantial body of data indicates that dopamine activity in the midbrain is tightly associated with reward processing. In humans, although indirect evidence from pharmacological and clinical studies also supports such an association, there has been no direct demonstration of a link between midbrain dopamine and reward-related neural response. Moreover, there are no in vivo data for alterations in this relationship in older humans. Here, by using 6-[¹⁸F]FluoroDOPA (FDOPA) positron emission tomography (PET) and event-related 3T functional magnetic resonance imaging (fMRI) in the same subjects, we directly demonstrate a link between midbrain dopamine synthesis and reward-related prefrontal activity in humans, show that healthy aging induces functional alterations in the reward system, and identify an age-related change in the direction of the relationship (from a positive to a negative correlation) between midbrain dopamine synthesis and prefrontal activity. These results indicate an age-dependent dopaminergic tuning mechanism for cortical reward processing and provide system-level information about alteration of a key neural circuit in healthy aging. Taken together, our findings provide an important characterization of the interactions between midbrain dopamine function and the reward system in healthy young humans and older subjects, and identify the changes in this regulatory circuit that accompany aging.     

From the Cover: Cognitive control in media multitaskers

Chronic media multitasking is quickly becoming ubiquitous, although processing multiple incoming streams of information is considered a challenge for human cognition. A series of experiments addressed whether there are systematic differences in information processing styles between chronically heavy and light media multitaskers. A trait media multitasking index was developed to identify groups of heavy and light media multitaskers. These two groups were then compared along established cognitive control dimensions. Results showed that heavy media multitaskers are more susceptible to interference from irrelevant environmental stimuli and from irrelevant representations in memory. This led to the surprising result that heavy media multitaskers performed worse on a test of task-switching ability, likely due to reduced ability to filter out interference from the irrelevant task set. These results demonstrate that media multitasking, a rapidly growing societal trend, is associated with a distinct approach to fundamental information processing.     

From the Cover: PNAS Plus: Persistent cannabis users show neuropsychological decline from childhood to midlife

Recent reports show that fewer adolescents believe that regular cannabis use is harmful to health. Concomitantly, adolescents are initiating cannabis use at younger ages, and more adolescents are using cannabis on a daily basis. The purpose of the present study was to test the association between persistent cannabis use and neuropsychological decline and determine whether decline is concentrated among adolescent-onset cannabis users. Participants were members of the Dunedin Study, a prospective study of a birth cohort of 1,037 individuals followed from birth (1972/1973) to age 38 y. Cannabis use was ascertained in interviews at ages 18, 21, 26, 32, and 38 y. Neuropsychological testing was conducted at age 13 y, before initiation of cannabis use, and again at age 38 y, after a pattern of persistent cannabis use had developed. Persistent cannabis use was associated with neuropsychological decline broadly across domains of functioning, even after controlling for years of education. Informants also reported noticing more cognitive problems for persistent cannabis users. Impairment was concentrated among adolescent-onset cannabis users, with more persistent use associated with greater decline. Further, cessation of cannabis use did not fully restore neuropsychological functioning among adolescent-onset cannabis users. Findings are suggestive of a neurotoxic effect of cannabis on the adolescent brain and highlight the importance of prevention and policy efforts targeting adolescents.     

From the Cover: Exploring the perceptual inabilities of Eurasian jays (Garrulus glandarius) using magic effects

In recent years, scientists have begun to use magic effects to investigate the blind spots in our attention and perception [G. Kuhn, Experiencing the Impossible: The Science of Magic (2019); S. Macknik, S. Martinez-Conde, S. Blakeslee, Sleights of Mind: What the Neuroscience of Magic Reveals about Our Everyday Deceptions (2010)]. Recently, we suggested that similar techniques could be transferred to nonhuman animal observers and that such an endeavor would provide insight into the inherent commonalities and discrepancies in attention and perception in human and nonhuman animals [E. Garcia-Pelegrin, A. K. Schnell, C. Wilkins, N. S. Clayton, Science 369, 1424–1426 (2020)]. Here, we performed three different magic effects (palming, French drop, and fast pass) to a sample of six Eurasian jays (Garrulus glandarius). These magic effects were specifically chosen as they utilize different cues and expectations that mislead the spectator into thinking one object has or has not been transferred from one hand to the other. Results from palming and French drop experiments suggest that Eurasian jays have different expectations from humans when observing some of these effects. Specifically, Eurasian jays were not deceived by effects that required them to expect an object to move between hands when observing human hand manipulations. However, similar to humans, Eurasian jays were misled by magic effects that utilize fast movements as a deceptive action. This study investigates how another taxon perceives the magician’s techniques of deception that commonly deceive humans.

Investigations into how magic effects exploit constraints on cognition in humans have recently sparked the interest of psychologists and neuroscientists alike (1–3). The success of most magic effects is dependent on their ability to take advantage of the perceptual and attentional shortcomings of the spectator. As such, the application of magic effects to investigate the mind can yield thought-provoking results, highlighting the elaborate deceptive qualities of magic and the perceptive and attentional blind spots that they exploit (4–8). Indeed, the recently coined “science of magic” (9) or NeuroMagic (5) offers psychologists and neuroscientists an exemplar tool to explore the constraints of the human mind. Recently, we suggested that magic effects could pose an interesting avenue to investigate whether nonhuman animals (henceforth “animals”) possess similar attentional and perceptual blind spots and cognitive roadblocks (10). Notice that magicians and comparative psychologists explore areas of cognition from different points of view: Comparative psychologists are ultimately interested in why and how diverse minds operate, with a focus on similarities and differences between species in capacity and constraints on cognition, whereas magicians aim to create the impossible by focusing on how the mind can be “fooled.” Consequently, the use of magic effects in comparative psychology provides an interesting methodological tool to explore how diverse species perceive the world around them, by focusing on their shared psychological constraints instead of their cognitive prowess.Corvids (large-brained birds in the crow family including jays, ravens, and magpies) have often been observed altering their caching behaviors to secure their caches from potential pilferers (11–13). These birds utilize intricate and highly elaborate cache protection tactics comparable to the deceptive strategies employed by magicians (10, 14). For example, corvids can cache food items discretely in among multiple bluff caching events, making it difficult for the observer to trace the real caching event (12). Moving an object in a series of quick motions to make it harder for the spectator to track is a common technique in magic. Moreover, corvids conceal items in their throat pouch, akin to a magician’s use of false pockets, and will manipulate food items within their beak similar to sleight-of-hand techniques performed by magicians (15). Given that corvids naturally employ behaviors that are similar to the tactics used in magic effects, these species may be exploiting similar perceptive and attentional constraints to the ones exploited by magicians.Here, we performed three different experiments using a sample of six Eurasian jays (Garrulus glandarius), in which we utilized three diverse magic methodologies that are typically used to mislead humans into thinking an object has been transferred from one hand to the other. We also tested the effectiveness of the methodologies on a sample of 80 human participants, who observed a subset of the same conditions that were presented to the jays. These methodologies (i.e., palming, French drop, and fast pass; Fig. 1) are an intrinsic part of most magic effects, in which their success heavily relies on not being noticed by the spectator. Magicians often use concealment techniques like palming and the French drop to mislead the audience into thinking the object of attention has been transferred from one hand to the other, when, in reality, the object remains concealed in the initial hand. For these tactics to successfully fool an observer, the spectator requires an expectation of the outcome of making certain hand movements. Such intrinsic knowledge will lead the spectator to unknowingly overlook the unusual components of the effect that might reveal foul play. These expectations seem to underpin the effects, thus prompting the naïve spectator to replace an altered sequence of events with their typical counterpart (16, 17). This, in conjunction with humans’ propensity to utilize acquired information from a previous experience to fill in the gaps they do not entirely perceive [i.e., amodal completion (18, 19)], will make these magic effects hard to distinguish from real transfers (20, 21). As such, when the magician operates the hand movement without enabling the object to transfer from one hand to the other, the spectator will assume the transfer has been completed, as this is the most likely outcome (22, 23). Thus, to be fooled by palming and French drop techniques, the spectator inherently needs some knowledge regarding the motions they are about to observe and their typical outcome, because, without this primordial expectation, there would be no preconceptions for the effect to rely on.Open in a separate windowFig. 1.Sequence of movements required to create (A) palm transfer, (B) French drop, and (C) fast pass. Image sources: ©GraphicsRF (worm); ©peart (hands); ©thruer (hands); ©sstocker (hands).Little is known about corvids’ preconceptions of human hand motions. Given that birds do not possess similar appendages, the interest lies in either the birds having the notion of such appendages due to either experience or evolutionary pressure, or the absence of such preconceptions even if the birds have extensive experience observing human hand motions. The third experiment (i.e., fast pass) might require neither of those principles as its modus operandi relies on the fast motion of the effect, thus relying on humans’ inability to spot the moment in which the object is quickly transferred from one hand to the other. Birds have different visual perception from human and nonhuman apes, possessing a much wider field of view with both binocular and monocular vision (24, 25). Consequently, whether our sample of jays is liable to similar techniques of deception could highlight convergent blind spots in attention and perception, which might make humans and corvids susceptible to similar deception tactics.     

Baby schema modulates the brain reward system in nulliparous women

Ethologist Konrad Lorenz defined the baby schema (“Kindchenschema”) as a set of infantile physical features, such as round face and big eyes, that is perceived as cute and motivates caretaking behavior in the human, with the evolutionary function of enhancing offspring survival. The neural basis of this fundamental altruistic instinct is not well understood. Prior studies reported a pattern of brain response to pictures of children, but did not dissociate the brain response to baby schema from the response to children. Using functional magnetic resonance imaging and controlled manipulation of the baby schema in infant faces, we found that baby schema activates the nucleus accumbens, a key structure of the mesocorticolimbic system mediating reward processing and appetitive motivation, in nulliparous women. Our findings suggest that engagement of the mesocorticolimbic system is the neurophysiologic mechanism by which baby schema promotes human caregiving, regardless of kinship.     

From the Cover: Individual variations lead to universal and cross-species patterns of social behavior

The duration of interaction events in a society is a fundamental measure of its collective nature and potentially reflects variability in individual behavior. Here we performed a high-throughput measurement of trophallaxis and face-to-face event durations experienced by a colony of honeybees over their entire lifetimes. The interaction time distribution is heavy-tailed, as previously reported for human face-to-face interactions. We developed a theory of pair interactions that takes into account individual variability and predicts the scaling behavior for both bee and extant human datasets. The individual variability of worker honeybees was nonzero but less than that of humans, possibly reflecting their greater genetic relatedness. Our work shows how individual differences can lead to universal patterns of behavior that transcend species and specific mechanisms for social interactions.

How individuals in a community interact with each other gives rise to collective emergent properties of the community (1–5). It reflects the individuals’ personal preference, social roles, the external environment, and other numerous factors applicable to specific context. The distribution of interevent times, or waiting time between two consecutive events, for temporal social networks has been much studied because of its relation to information or disease spreading (3, 4). It has been shown that the heavy tail in the interevent time distribution is due to a decision-based queuing process, in which some tasks are more prioritized than others (6, 7). In contrast, the distribution of contact duration, instead of the interevent time, and its connection to the nature of social interactions have not been studied as much.We have measured the duration of interactions among thousands of honeybees (Apis mellifera) in a hive, well-known eusocial insects that are easy to experimentally manipulate. Among many possible modes of honeybee social interaction, we focused on trophallaxis, which is mouth-to-mouth liquid food transfer. Trophallaxis occurs not only for feeding but also for communication (8, 9), making it a model system to study social interactions and collective effects (10, 11). To measure the interaction time, all of the honeybees in a colony were fitted with a barcode (12). A high-resolution machine vision camera imaged them at the rate of one frame per second. Then a customized algorithm detected each interaction event by analyzing the images and identified each bee, its position, and its orientation (12) (Materials and Methods).Note that all of the honeybee data used in this work were originally generated by the authors for separate studies not for the purpose of testing our theory discussed in this paper or even to acquire the data needed for the theory. We used all of the data available to us, which were the trophallaxis social network data acquired in 2013 and analyzed in ref. 12 (1164_2013, 1140_2013, 1138_2013, 1174_2013, and 1170_2013, which are trials 1 to 5 in ref. 12) and the trophallaxis social network data acquired in 2016 from colonies with partial treatment with Juvenile Hormone analogue (13, 14) (789_JHA_2016 and 757_JHA_2016) (see Materials and Methods for more detail on the colony preparation). Our theory works for all of them, which indicates its robustness.Fig. 1A shows that the distribution of honeybee interaction duration is heavy tailed. The exponents of the power law are –2.4, –2.3, –2.2, –2.2, –2.2, –2.7, –2.0, and –2.0 for each dataset listed in the legend from the top (1164_2013) to the bottom (757_JHA_2016). If every bee were the same and every interaction happened by chance, one might naively expect to see a peaked distribution such as a Gaussian. However, the observed heavy-tailed distribution suggests heterogeneity or variability among the population.Open in a separate windowFig. 1.(A) Duration of trophallaxis as well as F2F events that are not detected as trophallaxis follows a heavy-tailed distribution. The first number in the dataset name represents the number of observed honeybees in each colony, while the last number represents the year the experiment was performed. Note that the number of individuals in the last two datasets is actually smaller than what was originally observed (676 and 639 instead of 789 and 757, respectively) because we analyzed the data after applying an additional filter to exclude the time series during which the colonies were perturbed by the treatment of JHA, but we decided to keep the name of the datasets. The numbers of interactions used to make the plot are 302,221, 205,787, 191,795, 259,923, 329,170, 1,207,778, 136,529, and 115,965 in the order the datasets are listed in the legend. (B) Human face-to-face (F2F) interaction in various settings exhibits a heavy-tailed distribution. The numbers of interactions used are 10,677, 19,774, 14,037, 32,425, 9,865, 77,520, 26,039, 4,591, and 33,750 in the order the datasets are listed in the legend. In both A and B, error bars indicate the SE. In B, lower error bars for bins with count 1 could not be drawn in logarithmic scale because it extends to 0.In order to improve the statistical power of our analysis, we also examined F2F events, where honeybees were close and oriented toward each other but not actually engaging in trophallaxis (15). F2F events occur about an order of magnitude more frequently than trophallaxis. F2F events include undetected trophallaxis and possible antennation, but the nature of the honeybee interaction during F2F is not as well defined as trophallaxis. Nevertheless, being F2F is a necessary but not sufficient condition for trophallaxis, so as long as the distance apart is not larger than the length of a honeybee, it would be expected that F2F events scale similarly to trophallaxis events. Indeed, coarse data for temporal networks retain some statistics of the actual interaction including heavy-tailedness of the contact duration distribution (16) (also confirmed by our results in Fig. 1A; see 1166_F2F_2013).In our work, the detection of trophallaxis events is estimated using a two-step filtering process because we are not able to observe directly fluid transfer through the proboscis. Our detection scheme is not subject to tracking error generated by bees that are misoriented or have other individual variations in visibility. Our method, described in detail in our earlier work (12), first selects bee pairs of close enough distance and orientation toward each other. Then we apply a second filter, selecting the bee pairs that are physically connected by proboscis through image processing. Through this second filter, we can filter out the pairs with inaccurate tracking of position or orientation because those pairs would not be connected by the proboscis. Even if there were innately poorly tracked bees, the multiple layers of filtering minimize the detection error on the trophallaxis events. There is quantitative evidence that there is minimal tracking error impact on our results, even without the second filter. The F2F dataset concerns bee pairs that satisfy only the geometric constraint regarding distance and orientation but are not detected as being connected by the proboscis. This dataset has not filtered out the pairs with inaccurate position and orientation through the second filter. However, as shown below, this F2F dataset gives the same power law and scaling laws as the trophallaxis datasets. In other words, the statistics of the data are retained regardless of whether the data have gone through the second filter, which is consistent with the notion that inaccuracy due to positional tracking error was already negligible after the first filter. Our earlier work (12) showed by subsampling of the trophallaxis interaction data at different sampling rates that the statistics of the trophallaxis interaction are robust against subsampling and false negatives, with the statistics of the times between trophallaxis events being robust to detection errors. This strongly suggests that the statistics of the duration of the trophallaxis events will also be robust.We compared the honeybee data with human data recorded by the SocioPatterns collaboration (16–24) to explore whether there are universal patterns of social interaction. Fig. 1B shows that human F2F interaction time in various settings also exhibits a heavy-tailed distribution. The exponents of the power law are –2.4, –2.7, –2.9, –3.4, –2.7, –3.6, –2.5, –2.9, and –2.6 for each dataset listed in the legend from the top (Highschool_2011) to the bottom (Workplace_2015). Such similarity across different systems indicates an unexpected universality governing the interaction in social systems and suggests that a minimal model (25) should be able to capture the salient features of the interactions.To construct such a minimal model, we treat the social bond between bees as an effective particle. We focus on bees here, but the model is also applicable to humans. The bond is the edge in the usual network representation of social interactions. This effective particle has two states representing an interacting pair and a noninteracting pair. The particle jumps from one state to the other with a fixed rate ω in the reaction coordinate, as depicted in Fig. 2. Although jumping happens in both directions, we focus on the jumping from the interacting state to the noninteracting state because the interaction time is the waiting time for the first jump in that direction. The distribution of the first jump time f(t,ω) is obtained by multiplying the probability not to jump at each time step until time t with the probability density to jump at time t. The first part, the probability not to jump until t, is (1−ωdt)t/dt, where dt is the interval for a time step. Taking the limit of infinitesimal time steps yields limdt→0(1−ωdt)t/dt=e−ωt. The distribution of the first jump time is then given by f(t,ω)=ωe−ωt.Open in a separate windowFig. 2.Schematic picture for the theory of interaction durations. The dotted circular area indicates the neighborhood of the center bee. The neighbors, or the potential partners, inside the circle are connected with the center bee by a bond. The bond between two bees is represented by a particle in the reaction coordinate. This particle has two states representing an interacting pair and a noninteracting pair, and it changes its state by jumping over the energy barrier E with rate ω. Among the neighbors, the center bee is assumed to primarily interact with a bee with the highest barrier because they form the stablest pair.Some pairs tend to have longer interactions than others, and this is reflected in their value of ω. To take into account this variability or heterogeneity within the community, we integrate the jump time distribution for a fixed ω over the distribution of rates p(ω). To determine p(ω), we use the Kramers theory for escape over a potential barrier (26): the distribution of ω is related to the distribution of energy barrier heights E through ω=ω0e−E where ω0 is a constant (26). Here we use a dimensionless energy scale E, which is already normalized with possible fluctuations.We propose that the barrier height distribution p(E) follows the extreme value distribution for maxima. As illustrated in Fig. 2, a given bee has multiple candidate partners with which to interact. Each possible pair is associated with a certain barrier height. The pair with the highest barrier would spend more time together because it is more difficult for the particle to escape. This partner is thus interpreted as the most likeable and ends up forming a pair. Then the observed energy barrier E is the maximum among the neighbors. The distribution of the maximum is taken to be the Fisher–Tippett–Gumbel distribution for maxima (27, 28), which is appropriate if the parent distribution for barrier heights is localized, as seems reasonable. Then we express p(E)=(α+1)e−(α+1)Ee−e−(α+1)E where α is an undetermined parameter. For large E, p(E)∼e−(α+1)E. We take the large E limit because the heavy-tailed behavior is observed at large t. Using ω=ω0e−E, we find that p(ω)∼ωα for small ω, equivalent to large E. Combining this p(ω) with the exponential pair interaction time distribution, we getf(t)=∫dωp(ω)f(ω,t)∼t−(2+α)(1)as the interaction time distribution for the population. The power law form suggests that the assumption about the parent distribution for barrier heights is valid. More details of this calculation are provided in SI Appendix. The quantity α in the exponent connects the community interaction time distribution f(t) with the distribution of barrier heights p(E).We remark that a similar derivation for the heavy-tailed time distribution as shown in Eq. 1 arises in the theory for defect jumping (29) and a model of traps (30, 31) in glass. However, the interpretation of p(E) in social interactions is different from the analogue in disordered materials. Atoms in a glass successively hop over multiple energy barriers in a rough potential landscape, so the integration over p(E) is an average over the energy barriers experienced by one atom. On the other hand, our particles for the bond between a pair jump over one energy barrier to change their state. Thus, the integration over p(E) is an ensemble average over the population.Next we turn to verifications of the predictions of this theory. The simple theory predicts an exponential pair interaction time distribution. The quantile–quantile plots for pair interaction times (Fig. S1) suggest that the pair interaction time distributions for both honeybees and humans are better expressed by hyperexponential distributions, which are weighted sums of two exponential distributions. The theory is not affected by this additive modification, as discussed in SI Appendix. There are so many pairs in each colony that it is not practical to show the goodness of hyperexponential fit for each pair separately. Therefore, we devised a data collapse to show the fit of all pairs in one figure. Only the pairs that yield more than seven points of evaluation for the empirical cumulative distribution function (ECDF) were considered. The cumulative distribution function (CDF) for a hyperexponential function is Y(t)=1−ge−ω1t−(1−g)e−ω2t, where g is the weight and ω1 and ω2 are the rates for each exponential. We define a new variable z≡ω1t and rewrite the CDF as F(z,g,ω1,ω2)≡(1/g)1−Y−(1−g)e−(ω2/ω1)z =e−z, where Y is ECDF. Then the x axis only depends on one variable z. If the data are well fitted by this functional form, plotting F(z,g,ω1,ω2) against e−z should produce a cloud of data points aligning with the y=x reference line.Fig. 3 A and B show that most honeybee and human pair interaction times are well fitted to hyperexponential distributions. Fig. 3 A and B, Insets, show the fit of a pair to provide some intuition of the fitting process. The fitted CDF tends to deviate more at small e−z, or large t, because the CDF value of the fitting function approaches 1 for t→∞, while the ECDF value is 1 at the longest observed interaction time.Open in a separate windowFig. 3.(A) Pair interaction time distributions for honeybee pairs are fitted to hyperexponential functions and collapsed together. The numbers of pairs used to generate the plot are 197, 99, 328, 443, 561, 1,806, 46, and 20 in the order the datasets are listed in the legend. The pair interaction time distributions of these pairs were fitted. (Inset) Fitting ECDF of pair interaction times of a pair from 1164_2013. (B) The same plot as A but with human pairs. The numbers of pairs used are 37, 82, 59, 58, 39, 143, 98, 15, and 171 in the order the datasets are listed in the legend. (Inset) Fitting ECDF of pair interaction times of a pair from Highschool_2011. (C) Comparison between the scaling of interaction time distribution and mean pair interaction time distribution for 1164_2013. (D) The same plot as C but for Primaryschool. The number of mean pair interaction times used is the same as the number of pairs used for fitting, which is listed in A and B. The number of interaction times used is the same as what is listed in Fig. 1. Error bars indicate the SE, and lower error bars for bins with count 1 could not be drawn in logarithmic scale because it extends to 0.A second prediction from the model is the exponential barrier height distribution. Although E is not a directly measurable variable, the relation ω=ω0e−E enables us to indirectly measure p(E) because the mean pair time associated with an energy barrier is τ=1/ω. The relation p(E)∼e−(α+1)E implies that p(τ)∼τ−(2+α) for large τ, which has the same exponent as f(t) in Eq. 1. Therefore, comparing the exponent of the tail of f(t) and p(τ) provides a test of the theory, in particular, the proposed functional form of p(E).Fig. 3 C and D demonstrate the same scaling between the tail of f(t) and p(τ) for 1164_2013 and Primaryschool, respectively. The comparison of scaling for seven other honeybee datasets and eight other human datasets is shown in SI Appendix, Fig. S4. Here τ is obtained from fitting parameters ω, not from averaging of pair interaction times, because τ is the mean pair interaction time associated with a single energy barrier (SI Appendix). If we retain the full form for p(E), i.e., including the superexponential term in the Fisher–Tippett–Gumbel distribution, p(τ) is expected to have a peak at small τ, which may explain the peak in Fig. 3C.One might think that the identical scaling between f(t) and p(τ) is a consequence of the so-called stable law (32) because τ is an average of t for pairs. Then, depending on the tail of f(t), the distribution of τ would be given either by the central limit theorem, i.e., a Gaussian, or by a power law with the same exponent as f(t). However, this is not correct, as explained in SI Appendix, because the parent distribution of p(τ) is not f(t) but instead is the pair interaction time distribution f(ω,t).The two energy barriers suggested by the hyperexponential pair interaction time distribution imply a multidimensional potential landscape of the reaction coordinate. Our model does not limit the number of barriers, allowing the pair interaction time distribution in principle to be an arbitrary sum ∑igiωie−ωit, but the weight of further barriers seems to be too small to contribute to the dynamics. It is evident that different pairs are characterized by different barrier heights, but whether it is a specific pair or a specific individual that determines the barrier height cannot be determined by the analysis so far.To explore the effect of individuality in social interactions, we calculated the Gini coefficient (33) for 1) the total interaction time spent by each individual, 2) the total number of interactions each individual had, and 3) the total number of partners with which each individual interacted. Widely used to express inequality in economics, the Gini coefficients have recently been used to quantify inequality in the activity level of eusocial insects (34, 35). Fig. 4 shows a graphical representation of the results, known as the Lorenz plot (36), for the total interaction time spent by bees and humans. The Lorenz plots for other variables are shown in SI Appendix, Fig. S5.Open in a separate windowFig. 4.Lorenz plots of the total time spent for interaction by honeybees and humans. (A) Gini coefficients of bees are as follows: 1164_2013, 0.2373; 1140_2013, 0.2111; 1138_2013, 0.3013; 1174_2013, 0.2760; 1170_2013, 0.2698; 1166_F2F_2013, 0.2089; 789_JHA_2016, 0.1941; and 757_JHA_2016, 0.1727. The numbers of data points used in the plot are the same as the numbers of individuals in each dataset, which are 1,164, 1,140, 1,138, 1,174, 1,170, 1,166, 676, and 639. (B) Gini coefficients of humans are as follows: Highschool_2011, 0.4333; Highschool_2012, 0.4879; Hospital, 0.5488; Household, 0.5012; Hypertext, 0.4576; Primaryschool, 0.2799; SFHH, 0.4937; Workplace_2013, 0.4493; and Workplace_2015, 0.3753. The numbers of individuals used are 126, 180, 75, 75, 113, 242, 403, 92, and 217.To read these results, note that in a Lorenz plot the greater the deviation from the y=x reference line, the closer the Gini coefficient to unity, thus indicating a greater level of inequality. More inequality in our data means a greater contribution by individuals to the dynamics, signifying the effect of individuality. Fig. 4A shows that the Gini coefficients for honeybees are in the range 0.2 to 0.3, whereas for humans they are in the range 0.3 to 0.5. Thus, although individual bees are distinct, they are not as different from each other as humans are (Fig. 4B). The Lorenz plots and Gini coefficients for the total number of interactions and total number of partners provided in Fig. S5 show the same trend. The reduced individuality in honeybees compared to humans might be due to the average coefficient of relatedness being r=0.75 among workers of the same colony as the queen was inseminated with a single male in these experiments, but further study is needed to verify this conjecture. Since the interaction time is a shared value between a pair, it is nontrivial to completely separate the contribution of individuals. The effect of individuality in social interactions is therefore an open question but one that we have provided the tools to explore. Nevertheless, along with earlier studies of possible chemosensory mechanisms for individual identification (37), our results provide confirmation and quantification of the conjecture from recent work on the personality of honeybee workers as described in ref. 38 that some individuals are more likely to be interactive and engaged in food sharing, while others are less so.The recently discovered heterogeneous food distribution in the Camponotus sanctus ant colony (39) may suggest individual variations in workers of this other well-known eusocial insect. Although the ratio of transferred food volume to maximal transferable volume during trophallaxis when the donor is a forager is measured to follow the same exponential distribution with the same parameter as the case when the donor is a nonforager (39), it does not necessarily mean the lack of individuality in ants because the individual variations may have been averaged out as the data of many pairs were analyzed collectively. If the data were analyzed for each pair, individual variations may have been observed. It is not the scope of our work, but it would be possible to study the effect of individuality on the food mixing due to trophallaxis of eusocial insects.We have shown that high-resolution tracking can yield detailed multiscale information about the interactions and behavior of individuals within a community. Our results suggest that individual differences can lead to patterns of behavior that are universal and transcend species, context, and specific mechanisms for social interactions.     

From the Cover: Neural evidence that three dimensions organize mental state representation: Rationality,social impact,and valence

How do people understand the minds of others? Existing psychological theories have suggested a number of dimensions that perceivers could use to make sense of others’ internal mental states. However, it remains unclear which of these dimensions, if any, the brain spontaneously uses when we think about others. The present study used multivoxel pattern analysis (MVPA) of neuroimaging data to identify the primary organizing principles of social cognition. We derived four unique dimensions of mental state representation from existing psychological theories and used functional magnetic resonance imaging to test whether these dimensions organize the neural encoding of others’ mental states. MVPA revealed that three such dimensions could predict neural patterns within the medial prefrontal and parietal cortices, temporoparietal junction, and anterior temporal lobes during social thought: rationality, social impact, and valence. These results suggest that these dimensions serve as organizing principles for our understanding of other people.The human mind plays host to a panoply of thoughts, feelings, intentions, and impressions. External observers can never directly perceive these mental states—one can never see “nostalgia” nor touch “awe.” Nevertheless, humans are quite adept at representing other people’s internal states. Our ability to perceive and distinguish among the rich set of others’ mental states serves as the bedrock of human social life. We understand the fine differences between pure joy and schadenfreude and judge a friend’s glee accordingly. Our ability to distinguish a partner’s sympathy from sarcasm can make a world of difference to a relationship. Legal decisions frequently hinge on nuanced mental distinctions such as that between inattention and intentional neglect. How do people navigate such complexities in others’ internal mental worlds?One crucial tool for any navigator is a compass: a set of dimensions that help organize the contents of the world. By attending to the position of others’ mental states on key dimensions, humans might reduce the complexity of others’ minds to just a few essential elements—coordinates on a map. Might navigators of the world of mental states make use of such an intuitive compass? Research in other domains of cognition suggests such organization might be possible: The brain has a demonstrated capacity for extracting and capitalizing on useful regularities in the world. For example, our object representation system makes use of dimensions such as size and animacy to organize its processing tracts (1). Here, we explore the possibility that similar principles may organize our representations of other people’s minds.Decades of research in social cognitive neuroscience, primarily using functional magnetic resonance imaging (fMRI), have already implicated a well-defined set of brain regions in the process of thinking about mental states: Thinking about the lives and minds of others reliably engages a network including the medial prefrontal cortex (MPFC), medial parietal cortex (MPC), temporoparietal junction (TPJ), superior temporal sulcus (STS), and the anterior temporal lobe (ATL) (for a review, see refs. 2 and 3). However, this relatively young field has yet to explain how the social brain’s hardware processes the richness and complexity of others’ mental states. Fortunately, research in psychology supplies a set of theories regarding how people might organize their knowledge of mental states. The dimensions of these theories include valence and arousal (4, 5), warmth and competence (6, 7), agency and experience (8), emotion and reason, mind and body (9), social and nonsocial (2, 10, 11), and uniquely human and shared with animals (12). Any of these dimensions might plausibly play a role in organizing our understanding of mental states. But which, if any, do we spontaneously use during mentalizing? If a dimension actually matters to the way people typically think about others’ mental states, we should see evidence that the brain organizes its activity around that dimension. However, merely locating where in the brain mental state processing occurs—as social neuroscience has done so well already—cannot tell us how these regions represent mental states.Fortunately, new analytic techniques in functional neuroimaging, under the umbrella of multivariate or multivoxel pattern analysis (MVPA), enable us to bridge these levels of analysis. MVPA examines activity in distributed sets of voxels, allowing for discrimination between stimuli by their associated patterns of activity even when absolute magnitudes of activity remain constant. In this study, we use the form of MVPA known as representational similarity analysis (13) to test which psychological dimensions organize people’s understanding of mental states. These analyses work by measuring the extent to which neural patterns of activity can be predicted from theories of representational organization. To illustrate, the dimension “arousal” would predict that “ecstasy” and “rage” are represented very similarly in the brain because both are similarly intense mental states. In contrast, the dimension “valence” would predict that “ecstasy” and “rage” are represented very differently in the brain because one state is very positive, whereas the other is very negative. Both predictions can be tested by measuring the extent to which patterns of neural activity elicited by thinking about a person in ecstasy are similar to those elicited by thinking about a person in a fit of rage. Each dimension makes thousands of predictions about the similarity of each mental state compared with each other mental state; representational similarity analysis allows us to assess the accuracy of all of these predictions simultaneously. Thus, we can test which psychological dimensions capture the way the brain encodes others’ mental states.     

From the Cover: From simple innate biases to complex visual concepts

Early in development, infants learn to solve visual problems that are highly challenging for current computational methods. We present a model that deals with two fundamental problems in which the gap between computational difficulty and infant learning is particularly striking: learning to recognize hands and learning to recognize gaze direction. The model is shown a stream of natural videos and learns without any supervision to detect human hands by appearance and by context, as well as direction of gaze, in complex natural scenes. The algorithm is guided by an empirically motivated innate mechanism—the detection of “mover” events in dynamic images, which are the events of a moving image region causing a stationary region to move or change after contact. Mover events provide an internal teaching signal, which is shown to be more effective than alternative cues and sufficient for the efficient acquisition of hand and gaze representations. The implications go beyond the specific tasks, by showing how domain-specific “proto concepts” can guide the system to acquire meaningful concepts, which are significant to the observer but statistically inconspicuous in the sensory input.     

From the Cover: Subsecond dopamine fluctuations in human striatum encode superposed error signals about actual and counterfactual reward

In the mammalian brain, dopamine is a critical neuromodulator whose actions underlie learning, decision-making, and behavioral control. Degeneration of dopamine neurons causes Parkinson’s disease, whereas dysregulation of dopamine signaling is believed to contribute to psychiatric conditions such as schizophrenia, addiction, and depression. Experiments in animal models suggest the hypothesis that dopamine release in human striatum encodes reward prediction errors (RPEs) (the difference between actual and expected outcomes) during ongoing decision-making. Blood oxygen level-dependent (BOLD) imaging experiments in humans support the idea that RPEs are tracked in the striatum; however, BOLD measurements cannot be used to infer the action of any one specific neurotransmitter. We monitored dopamine levels with subsecond temporal resolution in humans (n = 17) with Parkinson’s disease while they executed a sequential decision-making task. Participants placed bets and experienced monetary gains or losses. Dopamine fluctuations in the striatum fail to encode RPEs, as anticipated by a large body of work in model organisms. Instead, subsecond dopamine fluctuations encode an integration of RPEs with counterfactual prediction errors, the latter defined by how much better or worse the experienced outcome could have been. How dopamine fluctuations combine the actual and counterfactual is unknown. One possibility is that this process is the normal behavior of reward processing dopamine neurons, which previously had not been tested by experiments in animal models. Alternatively, this superposition of error terms may result from an additional yet-to-be-identified subclass of dopamine neurons.Dopamine is an essential neuromodulator whose presence is required for normal learning, decision-making, and behavioral control (1, 2) and whose absence or dysfunction is associated with a variety of disease states including Parkinson’s disease, schizophrenia, addiction, and depression (3–7). Experiments in animal models support the hypothesis that changes in dopamine release at target neural structures encode reward prediction errors (RPEs) (the difference between actual and expected outcomes) important for learning and value-based decision-making (1, 8–12). In support of this claim, direct recordings of spike activity in mesencephalic dopaminergic neurons in nonhuman primates demonstrate that these neurons encode prediction errors in future reward delivery (8–10, 13, 14) and they may also encode other computations relevant for reward-guided actions (1, 15–17). However, action potential production in brainstem dopaminergic neurons can only be part of the story because activity in parent axons must be converted to changes in neurotransmitter release at synaptic terminals to have any impact on downstream neural systems (1, 18). There have been no direct measurements of dopamine release in human striatum that tests these ideas directly. In a large cohort of human subjects (n = 17), we tested the hypothesis that subsecond fluctuations in dopamine delivery to the human striatum encode RPEs generated during a sequential choice task.Our measurements of dopamine release are made in patients undergoing deep brain stimulating (DBS)-electrode implantation for the treatment of Parkinson’s disease. This patient population provides a unique and important window of opportunity to investigate dopamine’s role in human brain function. Parkinson’s disease symptoms are treated with dopamine replacement therapies, and yet we know nothing about how rapid (subsecond) dopamine concentration changes contribute to their symptoms or changes in their decision-making abilities. The opportunity to measure dopamine release with subsecond temporal resolution in the brains of humans with Parkinson’s disease is an opportunity to learn about fundamental processes in human brain function as well as an opportunity to assess dopamine signaling in a patient population whose primary treatment is focused on replacing function lost as dopamine neurons degenerate.Participants (n = 17) in these experiments performed a simple, yet engaging, sequential investment game (Fig. 1 and refs. 19–21) while dopamine measurements with subsecond temporal resolution were made in the striatum (n = 14 in the caudate and n = 3 in the putamen). Participants were offered participation after they were deemed candidates for deep brain-stimulating electrode implantation (22, 23). The research protocol was explained to the participants verbally, and they were provided a written consent form, as required by dual-institutional review board (IRB)-approved protocols at Wake Forest University Health Sciences and Virginia Tech Carilion Research Institute. Patients thus indicated that they understood the research protocol and provided written informed consent to proceed with the research procedure.Open in a separate windowFig. 1.Investment game. (A) Participants played a sequential-choice game during surgery using button boxes (Left) and a visual display (Right). For each patient, bet size adjustments (e.g., increase bet or decrease bet) and the decision to submit one’s answer were performed with button boxes. (B) Investment game (19, 21): participants view a graphical depiction of the market price history (red trace), their current portfolio value (bottom left box), and their most recent outcome (bottom right box) to decide and submit investment decisions (bets) using a slider bar in 10% increments (bottom center). Bet sizes were limited to 0–100% (in 10% increments) of the participant’s portfolio—no shorting of the market was allowed. During an experiment, a participant played 6 markets with 20 decisions made per market. (C) Timeline of events during a single round of the investment game.The sequential investment game (Fig. 1 and refs. 19–21) consists of 120 investment decisions. On each trial (t), this game requires participants to use button boxes to adjust and submit an investment [bet (b_t), where bet sizes could range from 0% to 100% of the participants portfolio, in 10% increments], after which, participants experience a gain or loss (participant return) equal to the bet size times the fractional change in the market price [market return (r) at time t: rt= Δptpt, where p is the market price and the participant return (i.e., gain or loss) at time t is equal to b_tr_t]. Previous work used this task and functional magnetic resonance imaging to demonstrate that RPEs and CPEs over gains are tracked by blood oxygenation level-dependent (BOLD) responses in the striatum (19, 20). These reports also demonstrated at the behavioral level that humans use counterfactual information over choices that “might have been made” and RPE information over choices that were actually made to make their next choice (19, 20).     

From the Cover: Exploitative leaders incite intergroup warfare in a social mammal

Collective conflicts among humans are widespread, although often highly destructive. A classic explanation for the prevalence of such warfare in some human societies is leadership by self-serving individuals that reap the benefits of conflict while other members of society pay the costs. Here, we show that leadership of this kind can also explain the evolution of collective violence in certain animal societies. We first extend the classic hawk−dove model of the evolution of animal aggression to consider cases in which a subset of individuals within each group may initiate fights in which all group members become involved. We show that leadership of this kind, when combined with inequalities in the payoffs of fighting, can lead to the evolution of severe intergroup aggression, with negative consequences for population mean fitness. We test our model using long-term data from wild banded mongooses, a species characterized by frequent intergroup conflicts that have very different fitness consequences for male and female group members. The data show that aggressive encounters between groups are initiated by females, who gain fitness benefits from mating with extragroup males in the midst of battle, whereas the costs of fighting are borne chiefly by males. In line with the model predictions, the result is unusually severe levels of intergroup violence. Our findings suggest that the decoupling of leaders from the costs that they incite amplifies the destructive nature of intergroup conflict.

Humans are capable of astonishing feats of altruism and cooperation (1–3), but, at the same time, of violent and destructive conflicts (4–8). A key factor contributing to the latter may be that wars are often waged at the behest of leaders who do not share fully in the immediate risks of conflict, and stand to gain benefits in terms of resources and status that are not enjoyed by the majority of combatants (4, 9–11). Could such “warmongering” be a feature of animal conflicts too? Only recently have models of animal aggression begun to explore the impact of inequalities among combatants in collective conflict (12, 13), and the usual assumption of existing theory is that individuals who initiate intergroup conflicts also contribute most to group conflict effort and thereby confer fitness benefits on the rest of their group (a positive or “heroic” model of leadership) (14–17). Here, we explore the more sinister possibility that those who initiate conflict may actually harm their fellows in pursuit of their own interests by exposing them to the risks of conflict while contributing little to fighting themselves (a negative or “exploitative” model of leadership).     

From the Cover: Long-term social dynamics drive loss of function in pathogenic bacteria

Laboratory experiments show that social interactions between bacterial cells can drive evolutionary change at the population level, but significant challenges limit attempts to assess the relevance of these findings to natural populations, where selection pressures are unknown. We have increasingly sophisticated methods for monitoring phenotypic and genotypic dynamics in bacteria causing infectious disease, but in contrast, we lack evidence-based adaptive explanations for those changes. Evolutionary change during infection is often interpreted as host adaptation, but this assumption neglects to consider social dynamics shown to drive evolutionary change in vitro. We provide evidence to show that long-term behavioral dynamics observed in a pathogen are driven by selection to outcompete neighboring conspecific cells through social interactions. We find that Pseudomonas aeruginosa bacteria, causing lung infections in patients with cystic fibrosis, lose cooperative iron acquisition by siderophore production during infection. This loss could be caused by changes in iron availability in the lung, but surprisingly, we find that cells retain the ability to take up siderophores produced by conspecifics, even after they have lost the ability to synthesize siderophores. Only when cooperative producers are lost from the population is the receptor for uptake lost. This finding highlights the potential pitfalls of interpreting loss of function in pathogenic bacterial populations as evidence for trait redundancy in the host environment. More generally, we provide an example of how sequence analysis can be used to generate testable hypotheses about selection driving long-term phenotypic changes of pathogenic bacteria in situ.Some of the most important bacterial pathogens are opportunistic in the sense that they infect a compromised human host from the surrounding environment. In cases where such infections become persistent, the evolutionary changes accompanying the transition from the environment to the human body have been the subject of intensive research, and we now have some information on what distinguishes clinical from environmental isolates (1). Surprisingly, our understanding of how this process is driven by selection often remains speculative. To study bacterial cells, we must remove them from the host environment into the laboratory, which may release them from the selection pressures that we wish to understand.In parallel, progress has been made in understanding how bacterial populations respond to selection through in vitro experimental evolution. These studies show that phenotypic dynamics result not only in response to the environment but also, to social interactions as bacteria cooperate and compete with one another (2). Selection to outcompete neighbors can even lead to loss of traits that increase survival in the environment but are costly to produce (3–5). Such loss has been shown for a range of traits, such as extracellular enzymes, signaling molecules, and iron chelators (2). These exoproducts act as “public goods”: products that are beneficial to the group but vulnerable to exploitation by cheats that reap the benefit without paying the cost (6). Understanding selection on public goods is clinically relevant, because many are virulence factors (7, 8), and social interactions have also been shown experimentally to affect infection dynamics in vivo (9, 10).We investigate the importance of social interactions in infectious populations of Pseudomonas aeruginosa, which is both a model organism of social evolution research and the primary cause of chronic lung infection in patients with the genetic disorder cystic fibrosis (CF). CF patients usually acquire their first P. aeruginosa infection in childhood, and these infections can persist for years, despite antibiotic treatment (11). P. aeruginosa produces an iron-scavenging molecule, pyoverdine, that acts as a cooperative public good in vitro (12). Iron is essential for growth but bound to transferrin, heme, and hemoglobin in the human host (13). P. aeruginosa circumvents this by releasing pyoverdine, which binds to iron and is taken up by a specific receptor. Detection of pyoverdine and expression of pyoverdine genes in sputum samples confirm that the pathway is active, and likely beneficial, in the CF lung environment (14, 15). However, cells that are deficient in production (i.e., potential cheaters) have also repeatedly been isolated from patients (16, 17). The pyoverdine metabolism is, therefore, an ideal system for testing whether social dynamics observed in the laboratory also occur in human hosts.Our aim is to identify selection pressures driving any changes that we observe in pyoverdine production in the lung. Pyoverdine production may be an adaptive response to acquire a limited nutrient. It may be lost, therefore, in response to availability of other iron sources (18–20). Alternatively, production may be lost from the population even if iron is limiting as a result of cooperator–cheat dynamics. Crucially, patterns of evolution of the pyoverdine system differ depending on whether adaptation to the human lung or social interactions drive selection (Fig. 1). If pyoverdine does not provide a growth benefit in the lung, the entire system will be redundant, including receptor function. In contrast, if pyoverdine production is lost because of cheating, receptor function will remain beneficial as long as extrinsic pyoverdine is available. Only when cheating is not possible does the receptor also become redundant. We can, therefore, distinguish between the two selection pressures by determining if and when receptor function is maintained in bacteria that have lost the ability to produce pyoverdine.Open in a separate windowFig. 1.The pyoverdine system. (Upper) The pyoverdine receptor FpvA spans the cell wall. In the absence of bound pyoverdine, the anti-σ factor FpvR inhibits the expression of σ-factors FpvI and PvdS. Pyoverdine acquires iron from transferrin. When ferripyoverdine binds to the receptor, FpvR releases FpvI and PvdS. Release of FpvI initiates synthesis of the receptor FpvA, and PvdS initiates synthesis of pyoverdine (illustrated by arrows). (Lower Left) If pyoverdine production is lost as an adaptation to the lung, receptor function also becomes redundant, irrespective of whether pyoverdine produced by neighbors is available. (Lower Right) However, if pyoverdine production is lost because of cheating, we expect to see retention of receptor function in the presence of pyoverdine produced by others and function only lost in the absence of pyoverdine.Two Danish collections of genome-sequenced P. aeruginosa isolates provide the opportunity to study selection on pyoverdine metabolism in CF patients (Dataset S1). The first collection gives a detailed insight into changes occurring during the first 10 y of infection across 36 young CF patients with 54 different clone types (21), representing the transition from initial colonization to chronic infection. With frequent and extensive sampling from each patient (451 isolates; on average, 13 per patient), we can estimate the point of colonization of each clone type and thereby, the time period over which a given isolate has evolved. The second collection provides insight into the long-term dynamics of two clone types causing chronic infections, with samples from 24 adult patients (85 isolates) infected with the two Danish transmissible clone types DK1 and DK2 (22–24). The two transmissible clone types established and spread in the Danish CF patient group from 1973 and all of the older patients (who got chronically infected up to the beginning of the 1990s) harbor one or both of these. Afterward, segregation of patients in the clinic has largely eliminated transmission of these clone types. The DK1 and DK2 isolates, thus, typically come from now older CF patients who have each been sampled a few times, providing insight into the long-term dynamics but not on a fine scale at the early infection stage. For some of the analyses, as specified below, only the isolates from the young patients have been used.     

From the Cover: Cross-modal individual recognition in domestic horses (Equus caballus)

Individual recognition is considered a complex process and, although it is believed to be widespread across animal taxa, the cognitive mechanisms underlying this ability are poorly understood. An essential feature of individual recognition in humans is that it is cross-modal, allowing the matching of current sensory cues to identity with stored information about that specific individual from other modalities. Here, we use a cross-modal expectancy violation paradigm to provide a clear and systematic demonstration of cross-modal individual recognition in a nonhuman animal: the domestic horse. Subjects watched a herd member being led past them before the individual went of view, and a call from that or a different associate was played from a loudspeaker positioned close to the point of disappearance. When horses were shown one associate and then the call of a different associate was played, they responded more quickly and looked significantly longer in the direction of the call than when the call matched the herd member just seen, an indication that the incongruent combination violated their expectations. Thus, horses appear to possess a cross-modal representation of known individuals containing unique auditory and visual/olfactory information. Our paradigm could provide a powerful way to study individual recognition across a wide range of species.     

From the Cover: Aquatic biodiversity enhances multiple nutritional benefits to humans

Humanity depends on biodiversity for health, well-being, and a stable environment. As biodiversity change accelerates, we are still discovering the full range of consequences for human health and well-being. Here, we test the hypothesis—derived from biodiversity–ecosystem functioning theory—that species richness and ecological functional diversity allow seafood diets to fulfill multiple nutritional requirements, a condition necessary for human health. We analyzed a newly synthesized dataset of 7,245 observations of nutrient and contaminant concentrations in 801 aquatic animal taxa and found that species with different ecological traits have distinct and complementary micronutrient profiles but little difference in protein content. The same complementarity mechanisms that generate positive biodiversity effects on ecosystem functioning in terrestrial ecosystems also operate in seafood assemblages, allowing more diverse diets to yield increased nutritional benefits independent of total biomass consumed. Notably, nutritional metrics that capture multiple micronutrients and fatty acids essential for human well-being depend more strongly on biodiversity than common ecological measures of function such as productivity, typically reported for grasslands and forests. Furthermore, we found that increasing species richness did not increase the amount of protein in seafood diets and also increased concentrations of toxic metal contaminants in the diet. Seafood-derived micronutrients and fatty acids are important for human health and are a pillar of global food and nutrition security. By drawing upon biodiversity–ecosystem functioning theory, we demonstrate that ecological concepts of biodiversity can deepen our understanding of nature’s benefits to people and unite sustainability goals for biodiversity and human well-being.

Species losses and range shifts because of climate change, harvesting, and other human activities are altering aquatic biodiversity locally and globally (1–5). In aquatic ecosystems, not only are some species severely depleted because of overfishing or habitat loss (3, 6–8), the ecosystem-level dimensions of biodiversity such as the total number of species and their functional diversity have also changed (9). Beyond the loss of particular species, changes in ecosystem-level dimensions of biodiversity threaten numerous ecosystem services to humans, which include the cultural, economic, or health benefits people derive from nature (10–13). In many regions, such as tropical coastal systems, the cumulative impacts of human activities are severe and associated with strong declines in taxonomic and ecological functional diversity (6) and coincide with regions with a high dependence of people upon wild-caught seafood for food and nutrition (14). In temperate regions, where some coastal communities depend on local wild seafood harvests to meet their nutritional needs (15, 16), species richness may be increasing as species recover from exploitation and warmer oceans allow species to expand their ranges into new territory (1, 2, 17).There is growing concern that biodiversity change leads to changes in human health and well-being (10, 13, 18). Specific and quantitative links between aquatic biodiversity and human health that distinguish contributions of species diversity from those of biomass, as predicted by biodiversity–ecosystem functioning theory, have not been established. At a time of unprecedented global change and increasing reliance on seafood to meet nutritional demands (19), there is an urgent need to understand how changing aquatic ecosystem structure may alter the provisioning of seafood-derived human nutrition.Seafood, consisting of wild-caught marine and freshwater finfish and invertebrates, provides an important source of protein and calories to humans. Additionally, unlike staple foods such as rice or other grains, seafood can address multiple dimensions of food and nutritional security simultaneously by providing essential micronutrients, such as vitamins, minerals, and polyunsaturated essential fatty acids critical to human health (19–22). Given the multiple attributes of seafood that are valuable to human health, it is possible that the diversity of an aquatic assemblage, distinct from the inclusion of any particularly nutritious species, could support human well-being consistent with a large body of evidence for biodiversity’s major contributions to ecological functions (11, 23–26). Dietary diversity is a basic tenet of a nutritious diet (27) and it is widely appreciated that diets composed of more food groups and more species are more nutritious (28–31). Ecological measures of dietary diversity (diet diversity, species richness, functional diversity, and Simpson’s index of evenness) have been associated with the nutritional value of diets in a range of contexts (27, 29, 32–38). These studies rely on relationships between species included in the diet (or other food intake measures) and nutritional adequacy of reported diets. However, a simple correlation between dietary diversity and a measure of dietary benefits provides only partial support for a claim that biodiversity benefits human well-being, consistent with the same ecological processes by which biodiversity supports numerous ecosystem functions and services (23, 26). We build upon this foundation of empirical relationships between diet diversity and diet quality by placing this question in the quantitative ecological theoretical framework that relates biodiversity to function (24, 25), thereby laying the groundwork for additional development of links between biodiversity science and our understanding of human well-being.Ecological theory predicts that biodiversity can be ecologically and economically important, apart from the importance of total biomass or the presence of particular species (23, 39). According to theory and over 500 explicit experimental tests (23, 40, 41), diversity in ecological communities and agricultural systems enhances ecosystem functioning by two mechanisms: 1) more diverse assemblages may outperform less diverse assemblages of the same density or biomass of individuals because more diverse assemblages will include more of the possible species and are therefore more likely to include high-performing species, assuming random processes of including species from the species pool (a selection effect), or 2) more diverse assemblages of a given density (or biomass) contain species with complementary functional traits, allowing them to function more efficiently (a complementarity effect) (25, 39). For aquatic animals, increased diversity enhances productivity of fish biomass (42) and also enhances temporal stability of biomass production and total yields (43, 44), providing economic and nutritional benefits to humans related to increased stability of harvests and production of biomass for consumption (43). However, when considering aquatic species from the perspective of human nutrition, functions other than biomass production become relevant because total seafood biomass consumption is not predictive of micronutrient benefits from seafood (45, 46).Here, we test a hypothesis central to ecological theory in the 21st century: whether biodiversity per se (species richness and ecological functional diversity), distinct from the identities and abundance of species, enhances human well-being (Fig. 1). We chose a measure of human well-being distinct from provision of protein, calories, or total yields—the micronutrient and essential fatty acid benefits of seafood. For increasing biodiversity per se (as opposed to increasing total seafood consumption) to enhance nutritional benefits as predicted by biodiversity–ecosystem functioning theory (25, 47), the amounts of various nutrients within edible tissues must differ among species, and furthermore, nutrient concentrations must trade off among species, such that species that have relatively high concentrations of some nutrients also have relatively low concentrations of others (25). Specifically, a “biodiversity effect” (sensu ref. 25) on nutritional benefits requires that concentrations of multiple nutrients are negatively correlated with each other, or uncorrelated, when compared among species, creating a complementary distribution of nutrients across species. In contrast, if nutrient concentrations in edible tissue are positively correlated for multiple nutrients across species such that, for example, a species containing high amounts of iron also has a high essential fatty acid concentration, thereby containing multiple nutrients in high concentrations simultaneously, seafood species or ecological functional diversity in the diet would not be important. In the case of positive correlations among nutrient concentrations, the ecosystem service of nutritional benefits would be enhanced by consuming more fish biomass or by selecting a few highly nutritious species, without considering species richness or ecological functional diversity.Open in a separate windowFig. 1.Aquatic biodiversity increases human well-being because edible species have distinct and complementary multinutrient profiles (A) and differ in mean micro- and macronutrient content (shown here relative to 10 and 25% thresholds of recommended dietary allowance, RDA, guidelines) for representative finfish (Abramis brama, Mullus surmuletus), mollusc (Mytilus galloprovincialis), and crustacean species (Nephrops norvegicus). Biodiversity–ecosystem functioning theory predicts that nutritional benefits, including the number of nutrient RDA targets met per 100 g portion (NT; i, iii) and minimum portion size (P_min; ii, iv) (B and E), are enhanced with increasing seafood species richness. Orange dots in B and E correspond to potential diets of high and low biodiversity levels. Seafood consumers with limited access to seafood each day may not reach RDA targets if diets are low in diversity (D–F versus A–C; gray shading indicates proportion of population that meets nutrient requirements). DHA: docosahexaenoic acid, EPA: eicosapentaenoic acid.We aimed to bridge two distinct theoretical frameworks—the biodiversity–ecosystem functioning theory and human nutrition science—by quantitatively testing for effects of aquatic species richness and ecological functional diversity (48, 49) in seafood diets on nutritional benefits via complementarity or selection effects. We used the public health measure of recommended dietary allowance (RDA) index to quantify nutritional benefits. RDAs are nutrient-based reference values that indicate the average daily dietary intake level that is sufficient to meet the nutrient requirement of nearly all (97 to 98%) healthy individuals in a particular life stage and gender group (50). Here, we used the RDA for females aged 19 to 50 y (SI Appendix, Tables S1 and S2; see SI Appendix, Table S1 for definitions of key terms). We measured nutritional value in terms of concentrations relative to RDAs, and we refer to these recommended amounts (or portions thereof) as “RDA targets” (SI Appendix, Tables S1 and S2 and Metrics). We quantified nutritional value in two ways: 1) the minimum amount of seafood tissue (in grams) required to meet given RDA targets (either for a single nutrient or the five micronutrients and fatty acids simultaneously; referred to as “minimum portion size required,” P_min [SI Appendix, Table S1, Eq. A1, and Metrics]) and 2) the number of nutrients that meet an RDA target in a single 100 g seafood portion (NT, SI Appendix, Table S1, Eq. A2). By considering nutritional value per unit biomass in both metrics, we avoided confounding diversity of seafood consumed with the total amount consumed (Metrics). We first tested two hypotheses: 1) seafood species richness increases NT because of complementarity in nutrient concentrations among species, and 2) seafood species richness increases the nutritional value of a 100 g edible portion of seafood, thereby lowering the minimum portion size, P_min, and improving the efficiency with which seafood consumers reach nutritional targets (Fig. 1). Following biodiversity–ecosystem functioning theory, we predicted that increased species richness is correlated with ecological functional diversity (51) in potential seafood diets and that ecological functional diversity is related to diversity in the concentration of essential elements and fatty acids that have nutritional value to human consumers, such that species and ecological functional diversity yields increased nutritional benefits. We also tested the hypothesis that seafood diversity increases total intake of heavy metal contaminants because some aquatic animals are known to bioaccumulate toxic metals in their tissues. For this reason, variation in bioaccumulation among species could lead to a biodiversity effect on contaminant intake that is detrimental to human health.In a global analysis of over 5,040 observations of nutrient concentrations in 547 aquatic species (SI Appendix, Fig. S1), we considered the provision of nutritional benefits to human consumers. To assess whether the relationships between biodiversity and human nutrition benefits depend on the geographic extent (global or local) over which seafood are harvested or accessed (11), we tested whether seafood species richness is associated with higher nutritional value at local scales (versus global scale) in traditional Indigenous seafood diets in North America (SI Appendix, Methods 1.4). Seafood is critical for Indigenous groups, who on average consume seafood at a rate that is 15 times higher than the global average per capita consumption rate (16). To test our hypotheses at the geographic scale of local consumer communities, we complemented our global analysis with additional analyses of 25 to 57 species in 14 geographically constrained groups of species consumed together as part of traditional Indigenous diets (SI Appendix, Methods 1.4).     

From the Cover: In vivo imaging identifies temporal signature of D1 and D2 medium spiny neurons in cocaine reward

The reinforcing and rewarding properties of cocaine are attributed to its ability to increase dopaminergic transmission in nucleus accumbens (NAc). This action reinforces drug taking and seeking and leads to potent and long-lasting associations between the rewarding effects of the drug and the cues associated with its availability. The inability to extinguish these associations is a key factor contributing to relapse. Dopamine produces these effects by controlling the activity of two subpopulations of NAc medium spiny neurons (MSNs) that are defined by their predominant expression of either dopamine D1 or D2 receptors. Previous work has demonstrated that optogenetically stimulating D1 MSNs promotes reward, whereas stimulating D2 MSNs produces aversion. However, we still lack a clear understanding of how the endogenous activity of these cell types is affected by cocaine and encodes information that drives drug-associated behaviors. Using fiber photometry calcium imaging we define D1 MSNs as the specific population of cells in NAc that encodes information about drug associations and elucidate the temporal profile with which D1 activity is increased to drive drug seeking in response to contextual cues. Chronic cocaine exposure dysregulates these D1 signals to both prevent extinction and facilitate reinstatement of drug seeking to drive relapse. Directly manipulating these D1 signals using designer receptors exclusively activated by designer drugs prevents contextual associations. Together, these data elucidate the responses of D1- and D2-type MSNs in NAc to acute cocaine and during the formation of context–reward associations and define how prior cocaine exposure selectively dysregulates D1 signaling to drive relapse.Drug seeking associated with addiction is mediated in part by strong associations between the rewarding effects of drugs and the environment within which they are administered (1). The nucleus accumbens (NAc) is a key neural substrate of context–reward associations because it mediates multiple aspects of this process, including learning, selecting, and executing goal-oriented behaviors (2–5). Over time, contextual cues in the absence of drug can elicit neural responses that resemble that of the drug itself and trigger drug craving and seeking, resulting in relapse in drug-addicted individuals (1). Furthermore, an inability to extinguish previously formed associations is thought to contribute to pathological drug seeking that persists despite periods of abstinence (6–9). Thus, identifying the specific neuronal populations that drive these behaviors, and determining therapeutic strategies to manipulate the cell populations to promote extinction of these associations, would be advantageous to improving treatment outcomes in drug-addicted individuals.Increased activity of the mesolimbic dopamine system is a central mechanism underlying the reinforcing and rewarding actions of drugs of abuse, including cocaine, as well as the compulsive drug seeking that develops over time and characterizes an addicted state (10–12). Dopamine action in NAc is mediated predominantly via activation of D1 or D2 dopamine receptors that are expressed by largely nonoverlapping populations of medium spiny neurons (MSNs) (13). These two subtypes of MSNs exert opposite effects on behavior, with optogenetic activation of D1-type neurons promoting positive reinforcement and increasing the formation of cocaine reward–context associations and activation of D2-type neurons being aversive and decreasing cocaine reward (14, 15); related differences in behavioral responses are seen in response to D1 vs. D2 receptor agonists or antagonists (16). Changes in NAc MSN activity in vivo have been documented in response to acute and chronic cocaine (17–19); however, it has not been possible to obtain this information in a cell-type-specific manner.The aim of the current study is to overcome this major gap in knowledge and define changes in D1- and D2-type MSN activity in vivo driven by acute and chronic exposures to cocaine and to determine how these cell types respond during context–reward associations. Thus, whereas the optogenetic studies cited above define the consequences of D1 or D2 MSN stimulation on reward-related behaviors, how the cell types respond under physiological conditions remains unknown. Furthermore, because optogenetic stimulations are experimenter-defined they eliminate the temporally specific nature of the endogenous signaling that encodes information. Advances in Ca²⁺ imaging have made it possible to record from genetically distinct subpopulations of neurons in awake behaving animals (20, 21). Here, we combine Ca²⁺ imaging and fiber photometry with conditioned place preference (CPP) to establish the patterns of activity of D1- and D2-type MSNs in NAc during formation of reward–context associations and to determine how these patterns are dysregulated by prior chronic exposure to cocaine. We then causally establish the role of altered firing of a given cell type to key aspects of context–reward associations by use of chemogenetic approaches.     

From the Cover: Economic impacts of tipping points in the climate system

Climate scientists have long emphasized the importance of climate tipping points like thawing permafrost, ice sheet disintegration, and changes in atmospheric circulation. Yet, save for a few fragmented studies, climate economics has either ignored them or represented them in highly stylized ways. We provide unified estimates of the economic impacts of all eight climate tipping points covered in the economic literature so far using a meta-analytic integrated assessment model (IAM) with a modular structure. The model includes national-level climate damages from rising temperatures and sea levels for 180 countries, calibrated on detailed econometric evidence and simulation modeling. Collectively, climate tipping points increase the social cost of carbon (SCC) by ∼25% in our main specification. The distribution is positively skewed, however. We estimate an ∼10% chance of climate tipping points more than doubling the SCC. Accordingly, climate tipping points increase global economic risk. A spatial analysis shows that they increase economic losses almost everywhere. The tipping points with the largest effects are dissociation of ocean methane hydrates and thawing permafrost. Most of our numbers are probable underestimates, given that some tipping points, tipping point interactions, and impact channels have not been covered in the literature so far; however, our method of structural meta-analysis means that future modeling of climate tipping points can be integrated with relative ease, and we present a reduced-form tipping points damage function that could be incorporated in other IAMs.

Climate tipping points are subject to considerable scientific uncertainty in relation to their size, probability, and how they interact with each other (1–4). Their economic impacts are even more uncertain, and consequently, these are often ignored (5, 6) or given a highly stylized treatment that fails to accurately represent geophysical dynamics and is nearly impossible to calibrate (7–9). As a result, tipping points are only weakly reflected in the policy advice economists give on climate change, typically by way of caveats and contextualization, rather than an integral part of the modeling that gives rise to estimates of the social cost of carbon (SCC) and other economic metrics of interest.The very definition of climate tipping points has attracted significant scholarship (2, 9, 10). We associate them with perhaps the best-known definition of “tipping elements”: “subsystems of the Earth system that are at least subcontinental in scale and can be switched—under certain circumstances—into a qualitatively different state by small perturbations” (2). This is an intentionally broad and flexible definition that admits a variety of geophysical responses, including nonlinear feedbacks and both reversible and irreversible phase changes (9). This flexibility is important for our purposes because economic studies omit or inadequately capture geophysical processes of all these sorts. Adopting a narrower definition (for example, limited to abrupt, discontinuous changes) would lead us to exclude geophysical processes with large economic costs.     

From the Cover: Elephants can determine ethnicity,gender, and age from acoustic cues in human voices

Animals can accrue direct fitness benefits by accurately classifying predatory threat according to the species of predator and the magnitude of risk associated with an encounter. Human predators present a particularly interesting cognitive challenge, as it is typically the case that different human subgroups pose radically different levels of danger to animals living around them. Although a number of prey species have proved able to discriminate between certain human categories on the basis of visual and olfactory cues, vocalizations potentially provide a much richer source of information. We now use controlled playback experiments to investigate whether family groups of free-ranging African elephants (Loxodonta africana) in Amboseli National Park, Kenya can use acoustic characteristics of speech to make functionally relevant distinctions between human subcategories differing not only in ethnicity but also in sex and age. Our results demonstrate that elephants can reliably discriminate between two different ethnic groups that differ in the level of threat they represent, significantly increasing their probability of defensive bunching and investigative smelling following playbacks of Maasai voices. Moreover, these responses were specific to the sex and age of Maasai presented, with the voices of Maasai women and boys, subcategories that would generally pose little threat, significantly less likely to produce these behavioral responses. Considering the long history and often pervasive predatory threat associated with humans across the globe, it is likely that abilities to precisely identify dangerous subcategories of humans on the basis of subtle voice characteristics could have been selected for in other cognitively advanced animal species.The ability to recognize predators and assess the level of threat that they pose is a crucial cognitive skill for many wild animals that has very direct and obvious fitness consequences (1–5). Until recently, most research in this area focused on how a range of birds and mammals classify other animal predators, demonstrating complex abilities to differentiate between predators with different hunting styles and respond with appropriate escape tactics (2, 3, 6–8). However, for many wild populations, humans represent a significant predatory threat (5, 9, 10) and this threat is rapidly increasing as areas for wildlife decrease and human–animal conflict grows. Moreover, as it is typically the case that not all humans pose the same risk to prey species, distinguishing between different human subgroups to identify those associated with genuinely threatening situations could present a major cognitive challenge. The extent of behavioral flexibility that different species may exhibit in correctly classifying human predators—and the degree of sophistication possible in such abilities—is therefore of considerable interest.Most research on the abilities of animals to classify human predators has focused on discrimination through facial features or general differences in behavior and appearance (5, 11–13). This focus has demonstrated that a number of different species are able to use visual cues to distinguish between individual humans that present varying levels of threat (4, 5, 14–16). However, acoustic cues could potentially provide a more effective means of classifying human predators by virtue of enabling categories of particularly dangerous humans to be identified. Such cues have an advantage because they code information on sex and age as well as cultural divisions that may be associated with differing levels of predation risk. Furthermore, these cues are available when the predator is still out of sight, potentially providing an important early warning system. Until now, however, studies of animal responses to human voices have focused on demonstrating skills in recognizing individual humans (17–21) rather than investigating specific abilities to identify particular human subgroups that have functional relevance in the natural environment.African elephants present an ideal model for a study of this nature, as humans constitute their most significant predator other than lions (1) and different human subgroups present them with different threats. African elephants are also already known to make broad distinctions between human ethnic groups on the basis of visual and olfactory cues (15). In the Amboseli ecosystem in Kenya, Maasai pastoralists periodically come into conflict with elephants over access to water and grazing for their cattle, and this sometimes results in elephants being speared, particularly in retaliation when Maasai lives have been lost (15, 22). In contrast, Kamba men, with more agricultural lifestyles, do not typically pose a significant threat to elephants within the National Park, and where conflict occurs outside over crop raiding, this largely involves male rather than female elephants (see, for example, ref. 23). Previous research has demonstrated that elephant family groups exhibit greater fear-based reactions to the scent of garments previously worn by Maasai men than Kamba men and also show aggression to presentations of the red clothes that Maasai typically wear (15). However, experiments involving presentations of human artifacts are inevitably limited in the level of natural variation that they can realistically simulate, whereas playback of human voice stimuli offers the possibility of investigating abilities to make a much wider range of functionally important distinctions. The extent to which elephants can use human voice cues to determine not only ethnicity, but also finer-scaled differences in sex and age that can dramatically affect predation risk, is highly relevant not only for determining the cognitive abilities that underlie predator recognition but also for understanding the coevolution of humans and arguably their most cognitively advanced prey.We used controlled playback experiments to investigate whether elephant family groups in Amboseli National Park were able to make subtle distinctions between the varying levels of threat posed by different categories of human. Although the presence of Maasai typically represents the greatest threat, this is specific to Maasai men because Maasai women are not involved in elephant-spearing events (22). We were able to compare behavioral responses of 48 female family groups not only to a large sample of voice stimuli from adult Maasai men versus Kamba men saying “Look, look over there, a group of elephants is coming” in their own language, but also to Maasai men versus Maasai women giving this utterance. In this latter experiment we used both natural stimuli and stimuli that had been resynthesized to mimic sex differences while keeping other acoustic characteristics of the voice unchanged. In a final experiment, we contrasted elephant responses given to the voices of Maasai men versus Maasai boys (see Materials and Methods for details).     

From the Cover: Interplay of approximate planning strategies

Humans routinely formulate plans in domains so complex that even the most powerful computers are taxed. To do so, they seem to avail themselves of many strategies and heuristics that efficiently simplify, approximate, and hierarchically decompose hard tasks into simpler subtasks. Theoretical and cognitive research has revealed several such strategies; however, little is known about their establishment, interaction, and efficiency. Here, we use model-based behavioral analysis to provide a detailed examination of the performance of human subjects in a moderately deep planning task. We find that subjects exploit the structure of the domain to establish subgoals in a way that achieves a nearly maximal reduction in the cost of computing values of choices, but then combine partial searches with greedy local steps to solve subtasks, and maladaptively prune the decision trees of subtasks in a reflexive manner upon encountering salient losses. Subjects come idiosyncratically to favor particular sequences of actions to achieve subgoals, creating novel complex actions or “options.”Humans and other animals often face complex tasks and environments in which they have to plan and execute long sequences of appropriate actions to achieve distant goals. One can represent the space of future actions and outcomes as a tree; such trees grow inordinately (often exponentially) large as a function of the length of the sequence (i.e., the depth of the tree). Rather little is definitively known about how this computational complexity is addressed. Work in the fields of reinforcement learning and artificial intelligence has suggested a number of heuristics that we describe below, namely, hacking, hierarchies, hoarding, and habitization (1–4). Various tasks have been designed to highlight individual heuristics; though how subjects generate and combine them without clear instruction has not been well characterized (however, see refs. 5 and 6).We previously designed a moderately deep planning problem to elicit a specific heuristic, in this case hacking or pruning of the decision tree (4). However, the task contains many of the elements that make choosing appropriately tricky in general. Thus, we closely examined the nature of, and individual differences between, the performance of subjects, shedding light on the interaction of heuristics in the self-generation of adaptive control when faced with a complex planning problem.Subjects had to plan a path through a maze so as to maximize their cumulative earnings. On each trial, they were placed in a random state and were asked to plan to a depth of 3, 4, or 5 moves (Fig. 1 A and B). Because each depth involved a binary choice, planning to depths 3, 4, and 5 corresponded to choosing among a set of 8, 16, or 32 possible sequences. We previously found that the large immediate losses at particular branch points in the tree (the red transitions) encouraged subjects to eliminate possibly lucrative subbranches beneath those points (4). This corresponds to suboptimal pruning or “hacking” of the decision tree (Fig. 1C). The analyses presented below show that this was by no means the only strategy subjects used.Open in a separate windowFig. 1.Task. (A) Task display. On each trial, subjects saw six boxes. The bright box indicated the randomly chosen starting location. The number of moves to plan was displayed at the top. During the decision time of 9 s, subjects had to plan between three and five moves. Then, during the input time of 2.5 s, they had to enter their plan as a single sequence of right/left button presses in one go and without immediate feedback as to what state they were currently in or what rewards they had earned in the choice sequence so far. After the entire sequence had been entered, the chosen sequence and the rewards earned were displayed in the order in which they had been entered. Failure to enter a button press sequence of the right length in the given time resulted in a penalty of –200 pence. (B) Task structure. Subjects were placed in one of the six boxes (“states”) at the beginning of each trial and had to plan a path through the maze that maximized their total outcomes earned. From each state, two successor states could be reached deterministically by pressing either the right (dashed lines) or left (solid lines) key. For example, from state 1, state 4 could be reached by pressing left–left–right. Each transition resulted in a deterministic reward or loss. Red arrows, for instance, denote large salient losses of –70 points. The possible transitions were never displayed on screen. (C) Pruning. The decision tree faced by subjects for a depth-3 problem starting in state 3. When encountering one of the large losses (−70, red arrows in B) the search along that subtree is terminated. The blue parts of the tree would thereby not be evaluated and thus the cost of computation would be reduced. In this case, pruning leads to a suboptimal sequence appearing as being optimal. (D) Hierarchical fragmentation of the same problem. Rather than evaluating the entire depth-3 tree, a 2–1 fragmentation would first search the tree up to depth 2 (large green area), choose a depth-2 sequence (black arrow), and then search the remaining depth-1 tree (bottom right green area). The blue area of the tree is again not evaluated. Optimal choices in the fragmented tree may miss the overall optimal sequence, which in this case would be on the far left of the tree. If a subject emitted the sequence on the far right, this sequence would be more likely under the fragmentation 2–1 than under a nonfragmented tree of full depth 3. The effective “subgoal” corresponding to the target of the first fragment (the end state of the subsequence resulting from the first part of the fragmentation) is indicated by a red asterisk.Hierarchical task decomposition licenses strategies for reducing computational burdens based on divide and conquer (2, 7) or “chunking” (5, 8–12). The resulting divided problems are easier to conquer because chunks are smaller and ignore aspects of the environment that do not impinge on their domains. The solutions to the subproblems can then be treated as larger-scale actions, often called macroactions or options (1). These simplify solving complex tasks by providing a way of building large decision trees out of smaller numbers of intermediate-sized parts (the macroactions) rather than larger numbers of smaller parts (each individual action). The downside is potential suboptimality. We use the precise form of suboptimality that our subjects exhibited to argue that they hierarchically fragmented the planning problems: Deep problems were solved by concatenating solutions to sequences of shallower problems (e.g., greedily adopting as a depth-5 solution the best depth-3 solution followed by the best remaining depth-2 solution; Fig. 1D). We then asked a critical question that has eluded previous approaches to hierarchical control, namely the degree to which the fragmentation is actually computationally advantageous—is the benefit of divide and conquer appropriately realized?A third, “hoarding” heuristic is known as memoization. If subjects are repeatedly faced with the task of finding a good policy at a state, then rather than building and searching the decision tree each time it is sensible to recall a previous solution and use that. If the previous solution cannot be guaranteed to be correct, then storing and deciding among several past solutions could be wise. When such storage and recall are probabilistic, the heuristic is stochastic memoization. It has been most extensively investigated in computational linguistics (13–15), and recently imported into decision making (16). Hoarding and hierarchies interact closely: Stored solutions can exactly be considered as macroactions or chunks, and so stochastic memoization can be seen as an answer to another poorly explored question, namely, how hierarchical decompositions arise. In particular, we will see that different subjects fragmented the task in idiosyncratic ways, putatively based on the way that they memoized.We used both flexible and constrained statistical analyses to examine the use of these heuristics. For the constrained analyses, we stipulated a particular mathematical form for each cognitive process and implemented it in a model that, after its parameters had been fit, reported the likelihood of the subjects’ choices. More complex models should provide better fits and were penalized by computing integrated Bayesian information criterion (iBIC) scores, which approximate Bayes factors (4, 17). We tested models including and excluding particular cognitive processes. Those processes present in the model with the best iBIC score were taken as putatively present in subjects’ decision making. Importantly, this approach always tests the ability of the hypothesized set of cognitive processes to account for the entire dataset, rather than only hand-selected aspects of the data.     

From the Cover: The absence of reward induces inequity aversion in dogs

One crucial element for the evolution of cooperation may be the sensitivity to others'' efforts and payoffs compared with one''s own costs and gains. Inequity aversion is thought to be the driving force behind unselfish motivated punishment in humans constituting a powerful device for the enforcement of cooperation. Recent research indicates that non-human primates refuse to participate in cooperative problem-solving tasks if they witness a conspecific obtaining a more attractive reward for the same effort. However, little is known about non-primate species, although inequity aversion may also be expected in other cooperative species. Here, we investigated whether domestic dogs show sensitivity toward the inequity of rewards received for giving the paw to an experimenter on command in pairs of dogs. We found differences in dogs tested without food reward in the presence of a rewarded partner compared with both a baseline condition (both partners rewarded) and an asocial control situation (no reward, no partner), indicating that the presence of a rewarded partner matters. Furthermore, we showed that it was not the presence of the second dog but the fact that the partner received the food that was responsible for the change in the subjects'' behavior. In contrast to primate studies, dogs did not react to differences in the quality of food or effort. Our results suggest that species other than primates show at least a primitive version of inequity aversion, which may be a precursor of a more sophisticated sensitivity to efforts and payoffs of joint interactions.     

From the Cover: Precursors to morality in development as a complex interplay between neural,socioenvironmental, and behavioral facets

The nature and underpinnings of infants’ seemingly complex, third-party, social evaluations remain highly contentious. Theoretical perspectives oscillate between rich and lean interpretations of the same expressed preferences. Although some argue that infants and toddlers possess a “moral sense” based on core knowledge of the social world, others suggest that social evaluations are hierarchical in nature and the product of an integration of rudimentary general processes such as attention allocation and approach and avoidance. Moreover, these biologically prepared minds interact in social environments that include significant variation, which are likely to impact early social evaluations and behavior. The present study examined the neural underpinnings of and precursors to moral sensitivity in infants and toddlers (n = 73, ages 12–24 mo) through a series of interwoven measures, combining multiple levels of analysis including electrophysiological, eye-tracking, behavioral, and socioenvironmental. Continuous EEG and time-locked event-related potentials (ERPs) and gaze fixation were recorded while children watched characters engaging in prosocial and antisocial actions in two different tasks. All children demonstrated a neural differentiation in both spectral EEG power density modulations and time-locked ERPs when perceiving prosocial or antisocial agents. Time-locked neural differences predicted children’s preference for prosocial characters and were influenced by parental values regarding justice and fairness. Overall, this investigation casts light on the fundamental nature of moral cognition, including its underpinnings in general processes such as attention and approach–withdrawal, providing plausible mechanisms of early change and a foundation for forward movement in the field of developmental social neuroscience.A dramatic shift in the study of morality has occurred, moving away from dissociable notions of moral development toward more integrated theories. An explosion of empirical research in psychology, anthropology, biology, economics, and neuroscience has resulted in an attempt to more clearly define and investigate the concept of morality across domains. Work in these various academic disciplines suggests that human moral sensibility emerges from a complex social, emotional, and cognitive integration, shaped by cultural exposure, and can therefore be seen as a product of our biological, evolutionary, and cultural history, representing an important adaptive element for social cohesion and cooperation (1, 2).Among the most exciting findings is the accumulation of evidence for early emerging capacities for social evaluations in infants and toddlers, interpreted as precursors to complex moral cognition (3). As demonstrated through a variety of techniques such as preferential looking-time, violation-of-expectation tasks, and behavioral observations, children under 2 y of age appear to both act prosocially and prefer prosocial to antisocial others (4). For example, 3-mo-olds preferentially attend to a character who previously acted in a prosocial manner toward another (5), suggesting a bias toward those that “do good things.” By 6 mo of age, this visual preference is expanded to the realm of behaviors; infants not only selectively attend to prosocial agents but also selectively approach them over antisocial or neutral characters (3). Within the first year, infants’ preferences are sensitive to the mental states of both the agents and the recipients of prosocial and antisocial acts as well as to the context in which such behaviors occur (3, 6, 7).In the second year of life, children’s reactions to social stimuli evolve from personal affective arousal to actual behavior such as helping, sharing, and comforting. Children 14 to 18 mo old help by fetching objects that an experimenter seems to want but cannot reach (8). Children between the ages of 1 and 2 y of age comfort others who are in distress and may go so far as to give up their favorite objects to soothe another (9). Children 18 to 25 mo old exhibit more concern for the victim of a moral transgression than for the transgressor, even if the victim did not show any behavioral signs of distress (10), suggesting that toddlers do not simply react to others’ emotional displays but actively interpret the intentions and feeling states of others. Although there is obviously considerable development in prosocial and moral abilities between infancy and childhood, it is less clear what motivates this change. There is some evidence that an improvement in one prosocial ability may relate to increased abilities in other prosocial areas. For example, 15-mo-old infants who chose to share a toy they preferred (compared with a nonpreferred toy or no toy at all) with an experimenter also attended significantly longer to a third-party interaction in which the allocation of resources among conspecifics was unequal (11). However, another body of literature suggests that early prosocial abilities are not necessarily related (12).Taken together, precursors to moral evaluation and prosocial behavior appear very early in development. However, the nature of these propensities remains highly contentious (13–15). Interpreting preverbal infants and toddlers’ behaviors and preferences has historically fallen into two camps: one advocating for lean interpretations based on rudimentary abilities/computations (16) and the other for rich interpretations based on complex cognitive and social cognitive processes (17). In the domain of morality, some argue that infants and toddlers possess a “moral sense” based on core knowledge of the social world (18), and others suggest that social evaluations are hierarchical in nature and the product of an integration of rudimentary general processes such as attention and approach and avoidance (19). Moreover, the development of these biologically prepared minds is embedded in and dependent upon specific dynamic social environments that are highly variable (20).Establishing neurological methods within a developmental framework has the potential to provide a more comprehensive account of morality, bridging the gap between behaviors and their underlying cognitive mechanisms. Neuroscience research is critical to clarify the computational systems that mediate early social evaluations and behaviors, often considered a prerequisite for moral thought. For example, examining the spatiotemporal dynamics of the neural processing when young children view social interactions can help us to better understand the contribution of domain-general processes to early moral thought. Our current knowledge of the brain circuits involved in the development of moral cognition is based on a limited number of studies with young children using electroencephalography (21–23), functional MRI (24), and lesion studies (25). Due to the methodological constraints of most neuroimaging methods, no study has yet investigated the link between the online neural processing of the perception of prosocial and antisocial others and actual moral preferences and prosocial behaviors in infants and toddlers, as well as their link to parental values.Recently, preschool-age children were shown to have both automatic and controlled neural differentiations when perceiving third-party harm and help (22). Moreover, the later controlled differences were predictive of children’s own sharing behavior. The association between neural computations involved in the processing of perception of harm and actual behaviors in children is beginning to be established, yet little is known about the early emergence of simple evaluations, particularly using neuroscience investigations with infants and toddlers. Some preliminary evidence suggests that infant’s post hoc evaluations of previously helping and hindering characters are marked by midrange attentional event-related potential (ERP) differences (P400) (23). However, although these findings have begun to illuminate the complexity of the neural underpinnings of early social sociomoral evaluations and prosocial behaviors, it is unclear whether these two fields (social evaluations and behaviors) are related in their emergence and whether there is online neural differentiation to the processing of harmful and helpful actions of others.There is accumulating evidence for early social evaluation, particularly between helping and hindering agents (3), but the social, cognitive, and affective processes and their interaction behind these early evaluations need to be identified (1). For instance, cognitive processes are required to understand the goals of actions and representation of agency. However, social evaluations of help and harm are also associated with relatively automatic reactions in the observer, potentially triggering basic approach versus avoidance mechanisms (26).Research on children and adult’s emotion regulation and social competencies has consistently shown that frontal power density asymmetry (particularly in 5–8 Hz frequency band for children and 8–13 Hz band for adults) during rest and when viewing emotional stimuli is related to individual differences in emotion regulation, motivational processes, and social behavior (27). For instance, asymmetries in resting state power density of the left and right frontal cortex are differentially involved in the processing of the emotions of others and are also related to one’s own emotional reactivity and approach–withdrawal behavior (28). This body of research has shown that greater frontal power density in the left hemisphere (vs. right hemisphere) is related to approach behaviors and positive affect and greater frontal power density in the right hemisphere is related to withdrawal and negative affect (29). Resting state measures have also been used to predict individual differences in social cognitive and prosocial development. For instance, an infant’s frontal and temporal power density asymmetries during resting state, collected at 14 mo of age, independently predicted comforting and helping behaviors several months later (24 and 18 mo of age, respectively) (30). In studies of preschool representational theory of mind, resting state frontal power densities in the alpha frequency band source localized to the right temporal–parietal junction and dorsal medial prefrontal cortex predicted theory of mind competency (31). Taken together, these studies suggest resting state frontal power density is an early index of social cognitive and emotion development, both competencies crucial for the development of morality.Infant temperament, parental socialization, and childhood environment are early antecedents of emotional regulation (32). Moreover, the complex interaction of child dispositions and the quality of parent–child relationships have long-term impacts, including social and cognitive development in childhood, adolescence, and adulthood (33). Furthermore, there is a growing body of evidence in adults that dispositional sensitivity to justice and fairness modulate online neural response to the perception of interpersonal harm in regions of the prefrontal cortex involved in cognitive control and decision-making (34, 35). Therefore, it is possible that these dispositions in parents will shape children’s prosocial behavior and neural responses during third-party evaluations of social interactions. Such parent/child value transmission is not the sole product of social learning but is rather a complex Gene × Environment interaction (36).Several time-locked neural responses in infants and children have been associated with differences in early visual differentiation of stimuli and relatively automatic responding (Nc) and controlled cognitive processes (positive slow wave, PSW) (37, 38). Each of these waveforms serves as a proxy for domain-general mechanisms of controlled and automatic processing in the developing brain. Moral sensitivity, in later development, is actually a careful integration of both an early automatic emotional component and a later cognitive reappraisal of stimuli, which can be explored in the temporal mechanical investigations of neural processing (1, 39). By combining EEG with eye-tracking, behavioral measures, and parental and children’s dispositions, the present study investigated the mechanisms at play during third-party social evaluations of prosocial and antisocial behaviors and their developmental trajectories. Two paradigms were used to assess implicit moral evaluations. One relied on a modified version of the helper versus hinder task (social evaluation task, SET), where children witness a character attempting to climb a hill and another character approaching to either assist or prevent the first character (40, 41). The other task was an infant-friendly version of the Chicago moral sensitivity task (CMST), developed by Cowell and Decety (22), in which two characters are interacting in a variety of prosocial (e.g., sharing, helping) and antisocial (e.g., shoving, tripping, hitting) ways. Consistent with developmental research, it was hypothesized that even the youngest infants would differentiate between characters that helped another and characters that hindered another, as indexed by differences in frontal power density asymmetries during the SET, and several key ERP components in the CMST. Specifically, as previous investigations of EEG frontal asymmetry have related relatively higher left power density to right power density as indexing emotional withdrawal (42), infants’ left density in the SET was expected to be greater than the right density in the perception of hindering condition, where infants would withdraw from the negatively valenced stimuli. Conversely, as frontal EEG asymmetries greater for right than left are related to approach behaviors (28), increased power densities in right compared with left frontal areas were expected in the perception of helping condition. Greater densities in the right versus left frontal areas (approach to prosocial agent) and preferential looking to helping versus hindering characters (as indexed by eye-tracking measures) were expected to directly relate to the child’s preferential reaching for a character that previously helped versus hindered another.Moreover, consistent with studies in older children, infants’ early, automatic, and later controlled time-locked neural responses to the perception of social interactions of others were expected in the CMST (21, 22). Infants were hypothesized to show greater amplitudes for good actions than bad actions in the Nc component, a central negativity between 300 and 500 ms poststimulus that has been previously linked to automatic resource allocation. The youngest infants (12 mo) were not expected to show later controlled differences between the processing of good and bad actions (in the PSW, 600–1,000 ms), however with age and the development of more elaborate cognitive processes, older toddlers (18–24 mo old) were expected to show greater amplitudes in the PSW for the perception of good actions, than bad actions, akin to the late positive potential (LPP) differences seen in preschool children engaged in similar implicit moral evaluations (24). Furthermore, individual differences in the later component were anticipated to predict both children’s sharing propensities and preference for the prosocial character. Finally, parental values regarding fairness and justice were expected to be reflected in early and late time-locked neural differences elicited by the perception of interpersonal harm.     

From the Cover: Ant colonies outperform individuals when a sensory discrimination task is difficult but not when it is easy

“Collective intelligence” and “wisdom of crowds” refer to situations in which groups achieve more accurate perception and better decisions than solitary agents. Whether groups outperform individuals should depend on the kind of task and its difficulty, but the nature of this relationship remains unknown. Here we show that colonies of Temnothorax ants outperform individuals for a difficult perception task but that individuals do better than groups when the task is easy. Subjects were required to choose the better of two nest sites as the quality difference was varied. For small differences, colonies were more likely than isolated ants to choose the better site, but this relationship was reversed for large differences. We explain these results using a mathematical model, which shows that positive feedback between group members effectively integrates information and sharpens the discrimination of fine differences. When the task is easier the same positive feedback can lock the colony into a suboptimal choice. These results suggest the conditions under which crowds do or do not become wise.