The Value of Sharing Information: A Neural Account of Information Transmission

Abstract

Humans routinely share information with one another. What drives this behavior? We used neuroimaging to test an account of information selection and sharing that emphasizes inherent reward in self-reflection and connecting with other people. Participants underwent functional MRI while they considered personally reading and sharing New York Times articles. Activity in neural regions involved in positive valuation, self-related processing, and taking the perspective of others was significantly associated with decisions to select and share articles, and scaled with preferences to do so. Activity in all three sets of regions was greater when participants considered sharing articles with other people rather than selecting articles to read themselves. The findings suggest that people may consider value not only to themselves but also to others even when selecting news articles to consume personally. Further, sharing heightens activity in these pathways, in line with our proposal that humans derive value from self-reflection and connecting to others via sharing.

Humans routinely share information with one another. What drives this behavior? One account suggests that humans have evolved disproportionately large brains, in part to coordinate socially (Dunbar, 2008; Schoenemann, 2006). People learn better when they anticipate opportunities to share with others (Lieberman, 2012), and the brain’s so-called default mode facilitates efficient social judgments (Spunt, Meyer, & Lieberman, 2013). Thus, sharing may be inherently promoted by human biology (Tamir & Mitchell, 2012). Social-network platforms, on which users share billions of messages daily (Facebook, 2015; Twitter, 2012), also reflect the motivation to share. In the current study, we tested the notion that the human biology may have evolved to support the motivation to share, or coordinate socially, and obtained novel evidence that connecting with others through sharing activates brain systems implicated in reward, social relevance, and self-relevance. We focused on online news as one form of sharing that has the potential for widespread impact (Pew Research Center, 2010).

Neural Precursors of Sharing

Studies of information selection and sharing have relied primarily on characteristics of the content or on self-reported responses (Berger & Milkman, 2012; Botha & Reyneke, 2013; Kim, 2015; Lee & Ma, 2012). However, people may not have the ability or desire to objectively reflect on their thoughts and emotions to explain their behavior (Dijksterhuis, 2004; Schmitz & Johnson, 2007). Furthermore, self-reports do not allow assessment of cognitive processes in real time, at the moment that individuals consider selecting or sharing information, and this limits understanding of the cognitive underpinnings of selection and sharing.

To address these limitations, we used neuroimaging (functional MRI, or fMRI) of activity in regions of interest (ROIs) as participants considered whether they wanted to read (select) or share New York Times articles. We focused on three sets of neural ROIs implicated in subjective value, self-related processing, and social cognition, respectively (Fig. 1).

Fig. 1.

The functionally defined regions of interest (ROIs) used in the ROI analysis. The ROIs are the white areas outlined in black.

Subjective value

We tested the idea that selecting and sharing information may carry inherent value. A meta-analysis of 206 fMRI studies (Bartra, McGuire, & Kable, 2013) found that activity in regions of the ventral striatum (VS) and ventromedial prefrontal cortex (VMPFC) is associated with positive valuation. We examined whether these meta-analytically defined regions were preferentially activated as participants considered selecting and sharing news articles.

We also tested whether activity in these regions scaled with preference to select and share articles. The perceived utility of content influences people’s choice of content (Botha & Reyneke, 2013; Kim, 2015), and greater activity in neural regions implicated in processing subjective value is associated with higher enthusiasm for sharing messages (Falk, O’Donnell, & Lieberman, 2012) and disclosing information about oneself (Tamir & Mitchell, 2012).

Self-related processing

We also tested whether brain regions implicated in self-relevance are associated with selecting and sharing information. A meta-analysis of 25 studies (Murray, Schaer, & Debbané, 2012) found that regions of the medial prefrontal cortex (MPFC) and posterior cingulate cortex (PCC) were engaged when participants made judgments about self-relevance. We tested whether these regions were engaged while participants considered selecting articles to read themselves. Individuals consider personal relevance when deciding to engage with content (Botha & Reyneke, 2013), and are biased toward content consistent with their preexisting beliefs (Cappella, Kim, & Albarracín, 2015). We also examined whether these ROIs were activated while participants considered sharing articles. Self-enhancement is a key motivation for sharing information, and people find value in sharing self-relevant messages (Berger, 2014; De Angelis, Bonezzi, Peluso, Rucker, & Costabile, 2012; Lee & Ma, 2012; Wien & Olsen, 2014).

In addition, we tested whether activity in these regions scaled positively with preference to select and share articles. Greater activity in neural regions implicated in self-related processing is associated with higher enthusiasm to spread ideas (Falk et al., 2012), and articles that resonate more personally are more likely to be selected and shared (Berger, 2014).

Social cognition

Finally, we examined brain activity associated with considering the mental states of other people (social cognition). Our social-cognition ROIs consisted of portions of the VMPFC, middle medial prefrontal cortex (MMPFC), dorsal medial prefrontal cortex (DMPFC), precuneus (PC), bilateral temporoparietal junction (TPJ), and right superior temporal sulcus (rSTS). In a study with a large sample (N = 462), Dufour et al. (2013) found that these regions were engaged when participants considered other people’s beliefs. We tested whether these regions were engaged when participants considered selecting information to read themselves. People regularly incor-porate others’ recommendations when making decisions, and this process is reflected in the activation of brain regions similar to the ROIs we examined (Cascio, O’Donnell, Bayer, Tinney, & Falk, 2015). Furthermore, social-cognitive components within the brain’s default-mode network prime people’s minds for social judgments even at rest (Spunt et al., 2013), and anticipation of sharing may be a key motive for reading content. In addition, given that sharing information is inherently social and social interaction is a key driver of news sharing (De Angelis et al., 2012; Berger, 2014; Lee, Ma, & Goh, 2011), we examined whether these regions were actively engaged while participants considered sharing articles.

We also tested whether activity in these social-cognition regions scaled positively with preferences to select and share articles. Anticipation of interpersonal interaction is a key motivation behind sharing (Lee et al., 2011), and sharing information may lead to social reward, such as interpersonal bonding (Berger, 2014). Thus, we tested the notion that people prefer to select and share articles that evoke social thoughts.

Summary

Our approach allowed us to measure brain activity within ROIs associated with subjective value, self-related processing, and social cognition (Fig. 1) while participants made judgments about selecting and sharing news articles, in real time. We used these data to test our proposal that people’s decisions to select and share information are based, in part, on the inherent reward associated with connecting with other people, as well as on considerations of self- and social relevance.

Method

Participants

Forty-three participants (30 female) between the ages of 18 and 24 (M = 20.5, SD = 2.1) took part in this study. Our target sample size of 40 participants was predetermined on the basis of funding, but because of concerns with data quality, we collected data from 3 additional participants before any statistical analysis was performed. Data collection stopped when we reached the enrollment goal. Two participants were excluded from analysis because of data corruption. This left 41 participants for our analyses. All participants gave informed consent in accordance with the procedures of the institutional review board of the University of Pennsylvania. Participants also met standard fMRI eligibility criteria; for example, potential participants were excluded if they had metal in their body, were currently taking any psychiatric medications, had a history of psychiatric or neurological disorders, were currently pregnant, or had claustrophobia. Participants were also required to be right-handed.

Procedure

Participants completed a baseline screening, as well as a neuroimaging appointment. During the neuroimaging appointment, they completed a series of self-report surveys and were scanned using blood-oxygen-level-dependent (BOLD) fMRI while they completed two tasks. In the task of interest to the present investigation, participants read 80 news headlines and abstracts that were published online in the health section of the New York Times between July 2012 and February 2013; these stimuli were divided into two runs of 40 news headlines and abstracts each. To control for reading speed, we had participants listen to recordings of the headlines and abstracts (M = 10.2 s, range = 8–12 s, SD = 1.41 s) while they read them. Each headline and abstract was randomly assigned to one of four conditions, within a randomization scheme that treated article length as a blocking factor (i.e., to balance the length of articles across conditions): In the broadcast-sharing condition, participants were asked, “How likely would you be to share this article on your Facebook wall?” In the narrowcast-sharing condition, they were asked, “How likely would you be to share this article with Facebook Friend _____?” (the name of a specific friend was inserted in the blank). In the select-to-read condition, they were asked, “How likely would you be to read the article yourself?” Finally, in the content-recall (control) condition, they were asked to indicate their certainty of the article’s topic (“How sure are you that [age/nutrition/fitness/science/laws/well-being/cancer] is the topic of this article?”). Participants responded to the questions on Likert scales from 1 (very unlikely) to 5 (very likely; in the content-recall condition, 1 = certainly not and 5 = certainly yes), thereby indicating their preferences to select or share the articles or their certainty regarding the topics of the articles.

Each trial began with a 1.5-s orientation screen that indicated the trial’s condition. The participants then saw (and heard via headphones) an article headline and abstract for 8 to 12 s. This display was followed by a fixation screen with a randomly jittered duration (M = 1.5 s, range = 0.5–4.7 s, SD = 0.97 s). Participants then had 3 s to record their response on a 5-point rating scale. A fixation screen was then presented for an intertrial interval, also of jittered duration (M = 2.0 s, range = 1.0–4.7 s, SD = 0.96 s). In order to avoid issues of collinearity be-tween trials, we used Optseq2 software (Optseq2, 2006) to maximize design efficiency. We ran 100,000 Optseq simulations, twice per run, to determine the optimal jitter times between trials and between the reading and rating screens within trials. Figure 2 illustrates the task design.

Fig. 2.

Illustration of the trial sequence in the article task. Participants were first reminded of the condition of the trial (a trial in the select-to-read condition is shown here). Then they saw (and heard an audio recording of) an article headline and abstract. This was followed by a jittered intratrial interval (M = 1.5 s). Finally, they were given 3 s to respond to a question, which was determined by the condition of the trial.

fMRI image acquisition

Neuroimaging data were acquired using 3-T Siemens scanners.¹ Two functional runs were acquired for each participant (500 volumes per run). Functional images were recorded using a reverse spiral sequence (repetition time = 1,500 ms, echo time = 25 ms, flip angle = 70°, −30° tilt relative to the anterior commissure–posterior commissure line, 54 axial slices,² field of view = 200 mm, slice thickness = 3 mm; voxel size = 3.0 × 3.0 × 3.0 mm). High-resolution T1-weighted images (magnetization-prepared rapid-acquisition gradient echo, 160 slices, slice thickness = 0.9 × 0.9 × 1 mm) and T2-weighted images were used in place with the BOLD images for coregistration and normalization.

Imaging data analysis

Functional data were preprocessed and analyzed using Statistical Parametric Mapping (SPM) software (Version 8, Wellcome Department of Cognitive Neurology, Institute of Neurology, London, United Kingdom). To allow for the stabilization of the BOLD signal, we did not collect data from the first five volumes (7.5 s) of each run. Functional images were despiked using the 3dDespike program as implemented in the AFNI toolbox (Cox, 1996). Next, data were corrected for differences in the time of slice acquisition using sinc interpolation; the first slice served as the reference slice. Data were then spatially realigned to the first functional image. We then coregistered the functional and structural images using a two-stage procedure; a 6-parameter affine transformation was used in each stage. First, in-plane T1 images were registered to the mean functional image. Next, high-resolution T1 images were registered to the in-plane image. After co-registration, high-resolution structural images were segmented into gray matter, white matter, and cerebrospinal fluid to create a whole-brain mask for use in modeling. T1 images were normalized to the skull-stripped Montreal Neurological Institute (MNI) template (MNI152_T1_1mm_brain.nii) provided by the FMRIB Software Library (FSL, 2012). Finally, functional images were smoothed using a Gaussian kernel (8 mm full width at half maximum).

Task analysis

Data were modeled using the general linear model as implemented in SPM8. Three conditions were modeled. The first condition (share) combined the two types of sharing trials, broadcast sharing (share on Facebook wall) and narrowcast sharing (share with a friend). The second condition (select) consisted of trials on which participants considered whether to select the full articles to read them-selves. The third condition (content) included the trials on which participants were asked to recall the content of the article and served as a control condition. Low-frequency noise was removed using a high-pass filter (128 s). The following contrasts were created: share > content, share > select, and select > content. Percentage-signal-change scores were extracted from each contrast for each participant using the MarsBar toolkit for SPM (Brett, Anton, Valabregue, & Poline, 2002). Next, a random-effects model was computed for each contrast, averaging across participants. Two sets of additional, parallel models were run: a model controlling for reaction time (RT) on each trial and a model using only a subset of trials that were matched on RT across conditions.

In addition, we examined the relationship between brain activity and participants’ preference ratings in the select and share conditions. These fixed-effects models, implemented in SPM8, used the preference rating as a parametric modulator of neural activity during each trial, for each participant. Next, a random-effects model was computed for each analysis at the group level, averaging across participants.

ROI analysis

To investigate neural response during the consideration of selecting and sharing news articles, we conducted a series of analyses using neural activity extracted from the three sets of a priori ROIs described earlier: VS and VMPFC for subjective value processing (Bartra et al., 2013), MPFC and PCC for self-related processing (Murray et al., 2012), and VMPFC, MMPFC, DMPFC, PC, bilateral TPJ, and rSTS for social-cognitive processing (Dufour et al., 2013; see Fig. 1 for brain maps showing these regions). Parameter estimates representing percentage signal change for each of the contrasts were extracted and averaged across participants.

Whole-brain analysis

In addition, following our planned ROI analyses, we examined the results of exploratory whole-brain analyses to determine whether neural regions outside of our ROIs were associated with the main contrasts of interest (select > content, share > content, share > select), as well as whether activity in these regions in the select and share conditions was modulated by the subsequent ratings. For all whole-brain analyses reported, we used a threshold of p < .05, k > 20, corrected for family-wise error using SPM8.

Results

Neural correlates of selecting and sharing articles

Decisions to select

We first examined whether making decisions to select articles was associated with brain activity in our a priori sets of ROIs (select > content contrast). All three sets of ROIs were more strongly activated when participants were thinking about selecting an article for themselves than when they were asked to recall the main content of the article (Table 1, Fig. 3).³

Table 1.

Results of the Three Contrasts in the Three Sets of Regions of Interest (ROIs)

	Select > content			Share > content			Share > select
ROIs	t(40)	p	Mean parameter estimate	t(40)	p	Mean parameter estimate	t(40)	p	Mean parameter estimate
Subjective valuation	7.22	< .001	0.118 [0.085, 0.151]	12.69	< .001	0.158 [0.133, 0.184]	3.09	.004	0.040 [0.014, 0.067]
Self-related processing	7.26	< .001	0.143 [0.103, 0.183]	15.25	< .001	0.225 [0.195, 0.255]	5.02	< .001	0.082 [0.049, 0.115]
Social cognition	4.99	< .001	0.067 [0.040, 0.095]	9.41	< .001	0.104 [0.082, 0.127]	3.12	.003	0.037 [0.013, 0.061]

Note: Values in parentheses are 95% confidence intervals. Table S3 in the Supplemental Material available online presents the activations in the subregions of each set of ROIs.

Fig. 3.

Estimates of percentage signal change in the subjective-valuation, self-related-processing, and social-cognition regions of interest, separately for the select and share conditions. Activation in each of these conditions was measured in contrast to activation in the content condition. The sagittal and axial cuts of the brain represent the regions of interest (white areas outlined in black). Error bars represent 95% confidence intervals.

Decisions to share

Next, we examined whether making decisions to share articles was associated with brain activity in our a priori sets of ROIs (share > content). All three sets of ROIs were more strongly activated when participants were thinking about sharing an article with other people than when they were focusing on the content of the article (Table 1, Fig. 3).

Effects of sharing versus selecting

Although both decisions to select and decisions to share articles were associated with activity in our subjective-value, self-related-processing, and social-cognition ROIs, when activity was measured relative to activity in the control condition, we next directly compared activity in the select and share conditions (share > select) to determine whether activation was stronger in one condition than in the other. We observed greater activation in all three sets of ROIs during the share condition than during the select condition (Table 1).

RT robustness analyses

We compared differences in RT between all conditions of interest. Although participants were slower to make decisions during the content trials than during the select and share trials, all ROI results remained robust in analyses controlling for RT and in analyses of a subset of trials that were matched on RT across conditions (see Tables S5, S6, S7, and S8 in the Supplemental Material available online). These robustness analyses suggest that our results were not driven by differences in difficulty across the conditions.

Whole-brain analyses

The whole-brain analyses examined whether regions outside of our a priori ROIs were more active during the select and share trials than during the content trials (select > content, share > content) or were more active during the share trials than during the select trials (share > select). The results of these analyses confirmed the results of our ROI analyses (see Table 2 and Fig. 4).

Table 2.

Results of the Three Contrasts in the Whole-Brain Analysis

Contrast and region	MNI coordinates			Number of voxels (k)	t(41)
	x	y	z
Select > content
Medial and ventromedial prefrontal cortex (bilateral)	−9	59	4	1,363	9.63
Dorsomedial prefrontal cortex	−18	38	43	92	6.75
Temporoparietal junction (left)	−51	−64	34	180	7.69
Precuneus (left)	−9	−55	19	81	6.65
Inferior temporal gyrus	66	−10	−14	52	6.30
Middle temporal gyrus	−60	−10	−17	141	7.48
Share > content
Medial prefrontal cortex (bilateral)	−6	53	10	3,263	15.58
Precuneus (right)	−6	−55	25	935	12.56
Temporoparietal junction (right)	51	−61	25	233	8.52
Temporoparietal junction (left)	−54	−67	43	336	8.20
Middle temporal gyrus	−63	−7	−14	202	8.47
Insula (left)	−30	17	−14	34	7.74
Inferior temporal gyrus	63	−7	−17	129	7.61
Hippocampus	−27	−34	−11	23	6.10
Share > select
Precuneus (bilateral)	9	−61	28	385	8.50

Note: The table reports significant activations ( p < .05, corrected for family-wise error; minimum cluster size = 20 voxels). The t tests were conducted at peak coordinates. MNI = Montreal Neurological Institute.

Fig. 4.

Results of the whole-brain analysis. The color coding indicates regions where, from left to right, the select > content, share > content, and share > select contrasts revealed significant activations ( p < .05, corrected for family-wise error; minimum cluster size = 20 voxels). See Table 2 for a detailed breakdown of the clusters, and see Figures S1 through S3 in the Supplemental Material for complete sets of sagittal slices illustrating these results.

Neural correlates of preference to select and share articles

Next, we examined whether activity in the neural regions in question scaled with participants’ degree of preference to select and share articles, respectively.

Preference ratings

On average, participants indicated that they had a higher likelihood to select articles (M = 3.17, SD = 1.40) than to share them (M = 2.12, SD = 1.26). Intraclass correlation (ICC) analyses revealed higher within-participants than between-participants variance in the preference ratings; individuals’ likelihood of selecting and sharing varied across articles, ICC1s = .18 and .20, respectively. In other words, individual participants expressed a range of preferences, rather than tending to rate all articles positively or negatively. Likewise, higher within-articles than between-articles variance in the preference ratings indicated that, across participants, the articles varied in their likelihood of being selected and shared, ICC1s = .11 and .07, respectively; thus, different participants preferred different articles, which suggests that the neural effects observed were not merely a function of article-specific features or due to some articles being universally preferred.

Neural correlates of likelihood to select and share

Activity in all three sets of ROIs was positively associated with higher preference ratings in both the select and the share conditions (see Table 3). We also conducted whole-brain analyses to more precisely identify neural subregions within and outside of our ROIs whose activation was associated with higher preference to select and share articles (see Table 4). The results of these whole-brain analyses supported the ROI analyses and suggested that the effects were relatively specific to our ROIs (i.e., we did not observe widespread activity outside of our main ROIs).

Table 3.

Results of the Region-of-Interest (ROI) Analysis Testing Modulation of Neural Activity by Preference Ratings

ROIs	Condition
	Select			Share
	t(40)	p	Mean parameter estimate	t(40)	p	Mean parameter estimate
Subjective valuation	6.01	< .001	0.046 [0.030, 0.061]	3.66	< .001	0.039 [0.017, 0.061]
Self-related processing	5.28	< .001	0.053 [0.033, 0.073]	3.36	.002	0.058 [0.023, 0.093]
Social cognition	3.47	.001	0.027 [0.011, 0.043]	3.20	.003	0.036 [0.013, 0.059]

Note: Values in brackets are 95% confidence intervals. Table S4 in the Supplemental Material available online presents the activations in the subregions of each ROI.

Table 4.

Results of the Whole-Brain Analysis Testing Modulation of Neural Activity by Preference Ratings

Condition and region	MNI coordinates			Number of voxels (k)	t(41)
	x	y	z
Select
Ventromedial prefrontal cortex (bilateral)	−6	38	−8	417	7.50
Cerebellum (right)	36	−61	−41	24	6.54
Middle temporal gyrus	−54	2	−23	47	6.54
Middle temporal gyrus	−63	−22	−14	50	6.31
Inferior frontal gyrus	−42	29	−2	41	6.07
Share
Dorsomedial prefrontal cortex	−12	53	34	48	6.31
Temporoparietal junction (left)	−48	−64	34	35	6.03
Middle frontal gyrus	−45	8	52	55	6.36
Caudate (right)	9	8	4	24	6.27
Caudate (left)	−9	14	7	29	5.91

Note: The table reports significant activations ( p < .05, corrected for family-wise error; minimum cluster size = 20 voxels). The t tests were conducted at peak coordinates. MNI = Montreal Neurological Institute.

Our whole-brain search did suggest, however, that a subportion of the VMPFC that is largely associated with self-related processing was associated with greater preference for selecting, but not sharing, articles. In contrast, subportions of the DMPFC and TPJ that are largely associated with social cognition were found to be associated with greater preference for sharing, but not selecting, articles (see Fig. 5).

Fig. 5.

Brain images showing regions where the whole-brain analysis indicated that neural activity in the select (left) and share (right) conditions was modulated by preference ratings.

Discussion

We propose that positive valuation, self-relevance, and social relevance drive people’s decisions to select and share information. Neural activity within subjective valuation, self-related-processing, and social-cognition ROIs was associated with deciding to select and share news articles, and scaled with preferences to do so. We observed substantial overlap in the processes underpinning selection and sharing decisions, though activity was heightened during sharing relative to selection. Scholars have previously suggested that similar psychological processes may underpin the selection and sharing of information (Cappella et al., 2015; Kim, 2015) and that representations of the self and other often overlap (Brewer, 1991; Platek, Keenan, Gallup, & Mohamed, 2004). Our data support these ideas by demonstrating neural overlap in the processes engaged.

Information selection

Our data are consistent with a value-based account of information selection; the VS and VMPFC are robustly associated with computing subjective values of stimuli (Bartra et al., 2013). One source of value for personal consumption of information may be an article’s self-relevance, and we observed greater activity within self-related-processing ROIs (MPFC and PCC) during decisions to select articles, relative to recalling the content. A second source of value may be an article’s social implications, and we observed greater activity within social-cognition ROIs (VMPFC, MMPFC, DMPFC, TPJ, PC, and rSTS) during decisions to select articles, relative to recalling the content. Activity in all three sets of ROIs was further associated with the degree of preference to select the article for oneself. These findings are consistent with previous literature on persuasion and influence (Cascio et al., 2015; Falk et al., 2012), which suggests that self-related processing may be a key factor in being influenced to act in accordance with a message (in this case, to select the article). These data are also consistent with the idea that even when selecting information for personal consumption, people may consider broader social factors (Cialdini & Trost, 1998). This finding converges with evidence that the default-mode network in the brain primes people to readily consider other people’s mental states (Spunt et al., 2013). Thus, social considerations may be important in selecting information, as the knowledge gained can translate into social value.

Information sharing

We also observed greater activity within all three sets of ROIs during decisions to share, both relative to decisions to select and relative to recall of content. Activity in all three sets of ROIs also scaled with the degree of preference to share the articles. These data are consistent with a value-based account of information sharing, in accordance with evidence that informing other people (Tamir, Zaki, & Mitchell, 2015) and sharing about oneself (Tamir & Mitchell, 2012) activates reward pathways. We extended these findings to the domain of sharing more broadly, and also examined two possible additional sources of value: self-relevance and social relevance.

Indeed, activity in meta-analytically defined self-related-processing regions of the MPFC and PCC was greater during sharing even when compared with making selections for oneself. These findings align with previous research demonstrating that MPFC activity scales with intentions to recommend ideas (Falk, Morelli, Welborn, Dambacher, & Lieberman, 2013). We extended these findings to show that merely considering sharing information activates this ROI, and the activity scales with preference.

These findings also highlight how the social act of sharing may be self-reflective, converging with accounts of self-presentation motives in sharing (Barasch & Berger, 2014). It has been suggested that desires to enhance one’s reputation and social status are key motivators behind news sharing (De Angelis et al., 2012; Berger, 2014; Lee & Ma, 2012; Wien & Olsen, 2014). Further, people are particularly likely to engage with messages that promote their values (Berger, 2014; Botha & Reyneke, 2013). Critically, our findings provide neural evidence that self-related processing is engaged not only when people consider selecting messages for themselves to read, but also when they consider sharing those messages with other people.

We also observed greater activity in our social-cognition ROIs in the share condition than in the select condition and the content condition. Also, this activity scaled with preferences to share. Humans have an inherent motivation to socialize through sharing information (Baumeister & Leary, 1995; Berger, 2014; Tamir & Mitchell, 2012). Prior research has shown that activity within subregions of the social-cognition ROIs is associated with successful retransmission of information (Falk et al., 2013) and enthusiastic recommendations (Falk et al., 2012).

Neural differences between selecting and sharing information

Although there was substantial overlap in neural activity when participants considered selecting and sharing information, the activity in all three sets of ROIs was strongest during decisions to share. In addition, we found preliminary support for some spatial distinctions in the areas engaged by preferences to select information to read oneself and preferences to share with other people. Specifically, our whole-brain results showed that a more ventral subportion of MPFC previously implicated in self-related processing and value to self was robustly associated with greater preference for selecting, but not sharing, articles. In contrast, DMPFC and TPJ areas previously implicated in social cognition were associated with greater preference for sharing, but not selecting, articles. These results support the proposed ventral-dorsal gradient of self- and other-related processing in the MPFC (Denny, Kober, Wager, & Ochsner, 2012) and suggest that although there is overlap of self-related and social-cognition activity in the selection and sharing of information, some specificity may also be involved when people consider how much they would like to read information as opposed to how much they would like to share it with other people.

In summary, we have proposed a novel account of the neurocognitive mechanisms behind selection and retransmission processes as participants actively consider selecting and sharing news. Increased activity in hypothesized subjective value, self-related-processing, and social-cognition ROIs was associated with decisions to select and share information, as well as with preferences to do so. These results suggest fundamental dimensions of the motivation to communicate and highlight more generally the overlap in processes involved in considering information for personal and social purposes.