Brain and Law: An EEG Study of How We Decide or Not to Implement a Law

Brazil has introduced a referendum regarding the prohibition of firearm commerce and propaganda arguments has invoked socially and personally driven issues in the promotion of voting in favor of and against firearm control, respectively. Here, we used different techniques to study the brain activity associated with a voter’s perception of the truthfulness of these arguments and their influence on voting decisions. Low Resolution Tomography was used to identify the possible different sets of neurons activated in the analysis of the different types of propaganda. Linear correlation was used to calculate the amount information <i>H(e<sub>l</sub>)</i> provided by different electrodes about how these sets of neurons enroll themselves to carry out this cognitive analysis. The results clearly showed that vote decision was not influenced by arguments that were introduced by propaganda, which were typically driven by specific social or self-interest motives. However, different neural circuits were identified in the analysis of each type of propaganda argument, independently of the declared vote (for or against the control) intention.


Introduction
As discussed before [1]: Any living being uses all possible resources to keep maintaining its identity in space and time. This means that any living being has as primordial goals to obtain not only the resources to synthesize their constituent chemicals, to obtain energy to support the chemical transactions and to allow it to creating other similar living beings, but also to achieve some goals dependent on some more abstract matters. Because of this, living beings have sets of goals (values) to attain and as resources on Earth are finite, this creates needs for goods and services to support FSV. Needs in turn motivates the individual to act in order to satisfy them.
Man is a social being and living together generates conflicts of interests triggered by opposing personal and social needs. Morality and law are aimed to offer solutions to solve or avoid these conflicts, providing two distinct sets of rules or norms for such a purpose. Whenever rules are broken and wrongdoing is established, punishment has to be provided and enforced to avoid future norm breaking. Wrong doing is prevented by coercion. Moral enforcement is a matter for divine or community reproach; therefore the degree of coercion of moral rules is low. In case of law, enforcement is the matter of Justice and Government. Justice is in charge of judging wrong doing and prescribing punishment, while Government is in charge of enforcing punishment [2] [3].
Positive law is a man-made law that obliges or specifies an action establishing specific rights for an individual or group in order to prevent wrong doing that may cause damage for the individual or group [4] [5]. To create a positive law is, therefore, necessary to identify a damage avoided and define an action to prevent it. A good positive law is well justified by an important threat to the individual and/or group and proposes effective action to prevent this well identified menace (e.g. [4]).
From time to time, tragedies like Newtown, CN shooting trigger waves of heated and emotional debates about some sort of gun control are basically polarized at two opposite ends of a broad spectrum: personal liberties versus safety and firearm injury. Proposals for gun control laws vary widely from country to country; while some countries try to restrict the sales of highly destructive weapons, other countries try to restrict the sales of all types of weapons. Within this second group, Brazil has recently carried out a referendum regarding the prohibition of firearm commerce (FC).
Media propaganda ( Table 1) used arguments (PA) tried, either to convince voters about the treat to the individual and society of having a free firearm commerce (YA) in order to promote voting in favor (Yes) firearm commerce control, or to stress the inefficacy of this kind of control (NA) in order to promote voting against (No) it (see Table 1). If PA was believed and NA discredited, firearm commerce control would be considered as an efficient way to prevent the damage caused by firearm house accident, traffic dispute, mass killing, etc. On the contrary if NA was not believed or PA discredited, firearm control should be ruled out in the referendum. However, to this true, voters would have to consider that propaganda influenced their voting.
YA and NA have different ontological origins. YAs are based on empirical facts because any threat is about incurred damage (see Table 1). Because of this, YA analysis requires recalling information from episodic memory. In contrast, NAs are supported by reasoning about personal versus collective rights, causes of criminality, etc. (see Table 1). Because of this, NA analysis requires using neural circuits supporting analytical reasoning.
The possible influence of a propaganda argument on vote decision depends on the supposed social threat being considered as real and important in YA case, or the alleged proposition being considered as right and implying and important personal loss in NA case. The threat veracity ( Table 1) demands its probability (p) being higher than zero, whereas its importance directly depends on p and the possible damage. Logical inference determines proposition veracity (Table 1) and the probability of failing to implement an adequate firearm control as well as the expected personal damage due to this failure is suggested to be high if the proposition is considered to be true. In this line of reasoning, highly probable believed YA would justify voting for firearm commerce control, while highly accepted NA would justify voting against the control. In order to influence vote decision, any PA has to justify one type of voting, but as each voter may have their own reasons (R) for voting Yes or No, even a justified PA may be of no influence if it does not match one of these reasons or it is not motivated by media propaganda to be included in R.
Rocha et al. [6] employed Principal Component Analysis (PCA) to study the amount of information ( ) i H e provided by electrode i e about vote intention one week prior to Election Day and showed that brain activity differed for "Yes" and "No" voters. "Yes" voters recruited neural circuits that have been previously described [7] in the Theory of Mind (ToM), which are involved in mentalizing third party intentions and important in the analyses of social issues. In contrast, "No" voters recruited neural circuits that have been shown to be involved

First opinion
Second opinion Here, you are asked to provide your opinion about some statements that have been made in free propaganda on the radio and TV.
Please select one of the following options that best describes your opinion about this statement: Please select one of the following options to evaluate the if this statement justify voting in the referendum: a) It justifies b) It does justify c) I do not know Please select one of the following options to evaluate the influence of this statement on your vote on Election Day: a) It will influence my vote b) It will not influence my vote c) I do not know with self-awareness and self-related processing [7]- [10]. In this context and from the above discussion, the following questions arise: 1) Are different neural circuits involved in YA and NA analysis because each type of argumentation has different ontological origins?
2) Is YA analysis mostly supported by episodic memory and social centered neural circuits and NA analysis mostly supported by analytical and self-related processing neural circuits?
In their study, Rocha et al. [6] asked 1136 people to complete a poll opinion questionnaire (PQ) to evaluate how true media propaganda arguments (PA) were, how much PA justified the referendum and how much the PA would influence their vote. In addition, 32 randomly selected individuals had their EEG recorded during their PQ responses. Here we used Loreta technology [11] and PCA analysis [6] to study 3 different EEG epochs of 2 seconds prior to the moments when the truth (T), justification (J) and influence (J) evaluations were decided, in order to test the hypotheses raised by questions 1 and 2 above.

Poll Opinion
A total of 1136 people living in the Great São Paulo area, a region inhabited by approximately 16 million people during the week that preceded the National Fire Arm Referendum, answered a poll opinion questionnaire (Table  1) about voting intention and propaganda arguments in favor (Y) and against (N) firearm commerce prohibition. Table 2 provides information on the sample population. The questionnaire shown in Table 2  Note that, as we intended to study the correlation between arguments and voting intention, and not voting decision-making, we carried out the measurements one week before the moment of voting.
PCA analysis confirmed the distinction between Y and N arguments because responses to the poll questions were divided into distinct components, as shown in Table 3.

EEG Recordings
Sixteen females and 16 males ( Table 2), all Caucasians and aged from 20 to 57 years old, were selected to have their EEG recorded (electrodes were placed according to the 10/20 system with an impedance of less than 10 kohm; low bandpass filter (50 Hz); a sampling rate of 256 Hz and 10 bit resolution, and ear lobe reference) while responding to the PQ. The EEG was visually inspected for artifacts prior to its processing, and the events associated with a poor EEG (e.g., when eye movements could compromise the results of the regression analysis) were discarded.
Two networked personal computers were used in the present study: one for the EEG recording and the other to sequentially display the different arguments in Table 1, where the individual's evaluation of each argument's truth (Yes or No) was requested. If it did (Yes) or did not (No) justify the referendum and if it would (Yes) or would not (No) influence the individual's vote intention. The volunteers were allowed to take as much time as they needed to make their decisions as well as to indicate whether they were undecided about any of these questions. We recorded the following times, represented the time question about the argument voting influence was introduced and the time of the individual's decision, respectively. The following EEG epochs were obtained for the analysis: ; The value of Δ was set as 2 seconds to ensure the volunteers were allowed sufficient time to reason and to make a decision that did not compromise our analysis. Low Resolution Electromagnetic Tomography (Loreta), which was developed by Pascual-Marqui [11], was used to localize the potential EEG source generators l s . The corresponding T, J and I EEG epochs were averaged for each electrode into different files according to each argument type (Y or N), which generated the EEG averaged files T Y and T N ; J Y and J N ; and I Y and I N ; respectively. Each of these files consisted of the corresponding EEG averages that were calculated for each of the 20 electrodes used to record the associated electrical activity for each decision epoch. A grand average was calculated for each of these files, and the corresponding z-score determined the EEG times that were statistically significant for the Loreta Analysis. Only those EEG moments with a z-score greater than 1.961 (5% significance level) were selected for this analysis (Figure 1). Only the areas showing the best match for each selected EEG moment was assumed as potential EEG source generators l s . Loreta analyses provided anatomical information and the Brodmann area number to localize the source of the EEG activity that was recorded at each selected time.
The amount of information    [12]. The rationality is the following.
The Pearson correlation R is +1 in the case of a perfect positive (increasing) linear relationship (correlation), −1 in the case of a perfect decreasing (negative) linear relationship (anticorrelation), and some value between −1 and +1 in all other cases, indicating the degree of linear dependence between the variables. As it approaches zero there is less of a relationship (closer to uncorrelated). The closer the coefficient is to either −1 or +1, the stronger the correlation between the variables. In this context, the correlation strength r is equal to R . If data ( ) and it quantifies the mean informational equivalence from s . Principal Component Analysis (PCA) is a statistical tool to investigate patterns of covariation in a large number of variables and to determine if information may be condensed into small sets of these variables called principal components [12]. This transformation is defined in such a way that the first principal component is the one that accounts for as much of the variability in the data as possible, and each succeeding component in turn explain the subsequent amount of variance possible under the constraint that it be orthogonal to (i.e., uncorrelated with) the preceding components. Factorial mappings are proposed to represent the activity of the neural circuits enrolled in a cognitive task because they condensed the information provided by the electrodes sampling this neural activity. In this ways, factorial analysis does not map brain areas activated by a cognitive task, but provides information to disclose the activity of circuits composed by neurons distributed on different areas of the brain recruited by the cognitive task, because

Results
Poll opinion and EEG data analysis are described below.

Poll Opinion
The results obtained from the PQ analysis are shown in Table 4. All of the voters considered half of the "Yes" arguments (YA) and 76% of the "No" arguments (NA) as true. The PA was assumed to not justify the FC by more than 75% of the individuals, and more than 64% of responders declared that this would not influence their vote. The data did not significantly change if the Yes and No voters were considered separately. These results showed that people were more convinced of the NA and coherently assumed that the PA did not justify the FC. The majority of voters declared that their voting decision would be determined by other reasons than those proposed by the PA.

EEG Results
PCA and Loreta analyses for each EEG epoch and the comparison of the results obtained with each technique are discusses bellow.

Truth Evaluation
Grand averages (Figure 1) were clearly different for YA and NA, having 40 significant moments (Z score greater than 1.961) in NA case and 70 of these moments in YA case. These differences resulted in two distinct LS sets of identified Loreta sources associated with Truth evaluation of each type of argument (Figure 2). Because of this PCA mappings associated with Truth evaluation were clearly different for YA and NA (Figure 2). Loreta sources were mostly identified at BA 10, 11, 17, 18, 29 and 47 for both types of argument but the frequency these sources were located at these locations differed for NA compared to YA.
PCA disclosed the existence of 3 patterns of ( ) i H e covariation that explained more than 60% of data covariance. P1 pattern consisted of F7, T3 and P4 in the YA case and PZ and F3 in the NA case. e about the sets of neurons activated by the cognitive task being processed. Loreta analysis identifies these sets of activated neurons and PCA discloses those electrodes providing similar information about them. Since the electrical activity recorded by each i e depends on the distance from the electrode to the activated neural sources l s , the nearest l s to each electrode that significantly loaded in each PCA pattern are the most probable set of neurons contributing to distinguish these patterns. Figure 3 display the XY coordinates of nearest l s to each PCA pattern in the case of Truth evaluation and Table 5 identify these sources.    Table 5.
Loreta analysis and PCA comparison shows that pattern P 1 is correlated with l s located at BA 6, 9 and 10 in NA case and BA 7, 22, 40 and 47 in YA case. In addition, sources associated with P 2 and located at Middle  Grey background highlight common source locations.

Justification Evaluation
Grand averages (Figure 2) were clearly different for YA and NA, having 50 significant moments in NA case and 80 of these moments in YA case. These differences resulted in two distinct LS sets of identified Loreta sources associated with Truth evaluation of each type of argument (Figure 2). Because of this PCA mappings associated with Truth evaluation were clearly different for YA and NA (Figure 4). Loreta sources were mostly identified at BA 10, 11, 17, 18 and 47 for both types of argument but the frequency these sources were located at these locations differed for NA compared to YA.
PCA mappings associated with Justification evaluation were clearly different for YA and NA (Figure 4). The P1 pattern was reduced in electrode P4 for the YA case and in the occipital electrodes for the NA case. Pattern P2 was very similar for both YA and NA, consisting of bilateral frontal electrodes. Pattern 3 consisted of F3, C3, P3, O1, FZ, CZ, PZ, OZ, C4 and O2 responses in the YA case, and comprised the F3, C3, P3, CZ, PZ and C4 electrodes.
Loreta analysis and PCA comparison (Figure 5 and Table 6) shows that pattern P 1 is correlated with l s lo- Figure 5. PCA mappings and spatial location of those Loreta sources associated with each PCA pattern in case of Justification evaluation. The sources are those described in Table 6.

Influence Evaluation
Grand averages (Figure 1) were clearly different for YA and NA, having 35 significant moments in NA case and 56 of these moments in YA case. These differences resulted in two distinct LS sets of identified Loreta sources associated with Truth evaluation of each type of argument (Figure 6). Because of this PCA mappings associated with Truth evaluation were clearly different for YA and NA (Figure 2). Loreta sources were mostly identified at BA 7, 10, 11, 18 and 21 for both types of argument but the frequency these sources were located at these locations differed for NA compared to YA.
PCA mappings associated with Justification evaluation were clearly different for YA and NA (Figure 6). Pattern P1 consisted of F3, T5 and T6 in the YA case and occiptial electrodes in the NA case. Pattern P2 was very similar for both YA and NA, consisting of bilateral frontal electrodes. Pattern 3 consisted of P3, O1, FZ, CZ, PZ, OZ, C4, P4 and O2 in the YA case and the F3, C3, P3, CZ, PZ and C4 electrodes.
Loreta analysis and PCA comparison (Figure 7 and Table 7) shows that pattern P 1 is correlated with l s mostly located at BA 18 in NA case and BA 21 and 22 in YA case. In addition, sources associated with P 2 and located at Middle Frontal Gyrus (BA 10 and 11

Discussion
A total of 1136 people living in the Great São Paulo area were asked whether media propaganda would influence or not their vote decision in the Brazilian Firearm Commerce Referendum. As discussed in the introduction, in order to justify a given type o vote, an argument has to be true and high probable. Each voter has their own set of arguments or reasons R for voting for (Yes) or against (No) firearm commerce control that may be dependent on many causes and believes. Therefore, it does not necessarily follow that a justified media argument will influence vote decision. In this line of reasoning, any media campaign is aimed to manipulate the inclusion of as many justified arguments pro a vote decision as possible in R.
Poll opinion results showed that arguments (YA) for voting Yes was considered true by 50% of the volunteers and while 76% of them agreed with arguments (NA) for voting No, no matter their declared voting intention. Despite this, only 20% of the volunteers considered that media arguments would justify the intended vote. As discussed before, besides being true, an argument has to be high probable in order to justify the intended vote decision. Therefore, it may be said that media campaign failed in proving their arguments to probable enough to justify voting for or against firearm control. In line with these results, more than 60% of the volunteers declared that media propaganda would not influence their vote decision. Because for most of people, media arguments were not justified, media propaganda also failed in changing voter's set of reasons R. This was a remarkable finding because it shows that the vote decision was guided by reasons other than those that are usually discussed in a debate when any form of gun control gains national attention. The analysis of the EEGs recorded during the PQ responses shed some light on this subject.
Both Loreta analysis and PCA show that the same individual recruited different neural circuits during T, J and I epochs to analyze YA and NA irrespective of their vote intention about firearm commerce prohibition. YA analysis recruited larger set of neurons than those recruited by NA, indicating, perhaps that YA evaluation requires a reasoning more complex than that required by NA.
The core of YA arguments is an empirical fact that would characterize a social threat to be avoided by the firearm control. Voters would have to recall these facts from memory in order to evaluate their probability p and associated damage. If p > 0 then the argument is true and if p > 0 then it is frequent. Now, if probability and damage are high an important social threat is identified. In contrast, the core of NA arguments implies reasoning about personal versus collective rights, causes of criminality, etc. In this case, it is necessary first to establish if the argument true and then its expected personal benefit and risk have to be estimated. Now, if argument is true, expected personal benefit is low and risk is high a strong case is made for No voting. Because of this, it is expected that neural circuits involved in YA and NA analysis have to be different as observed here. This is in line with other propositions in the literature, e.g., that of "exploration-explotation" dilemma by Nathaniel et al. [13].
It is proposed here that PCA pattern P 3 it is associated with the memory recalling of the fact proposed by YA, because this pattern is associated with sources located at BA 9 and 10 in the frontal cortex [14]- [19], BA 18 and 19 in the posterior cortex [20] and BA 38 and 40 at the temporal cortex [14] [20] [21]. In addition, pattern P 2 it A. F. Rocha et al.
574 Figure 7. PCA mappings and spatial location of those Loreta sources associated with each PCA pattern in case of Influence evaluation. The sources are those described in Table 7. is associated mostly with sources located at the right frontal cortex (mostly BA 9, 10, 11 and 47) that has an important role in episodic memory handling as discussed by Henson et al. [22] and Lepage et al. [16].
Complementing the above hypothesis, it is assumed here that PCA pattern P 3 it is associated with the analytical reasoning involved in establishing the veracity of the fact proposed by NA, because this pattern is associated with sources located at BA 11 in the frontal cortex, BA 3 and 5 in the parietal cortex and BA 17 and 18 in the posterior cortex [22]- [25]. In addition, pattern P 2 it is associated mostly with sources located at the left frontal cortex (mostly BA 9, 10, 11, 46 and 47) that has in important role in analytical reasoning as proposed by Canessa et al. [22], Knauff et al. [23] and Mackey et al. [25].
Finally, it must be stressed that P 1 is the largest and most distinctive PCA pattern associate with truth evaluation of YA and NA as it was the most distinctive pattern between Yes and No voters [6].
From the above, it is proposed here that pattern P 2 disclosed those frontal executive neural circuits handling temporal-parietal-occipital circuits disclosed by pattern P 3 and in charge of data recalling in case of episodic memory and YA analysis and in charge of logically inferring the truth of NA. Rocha et al. [6] also considered P 2 as associated with executive frontal neural circuits handling those other neural circuits disclosed by P 3 and in charge of differentiating Yes from No voters.
YA arguments were socially driven, which was typical for arguments in favor of any gun control in a variety of societies. ToM refers to the ability to attribute mental states to others and the ability to infer the emotional experiences of others (empathy), which are important processes in social cognition. Brain imaging studies in healthy subjects have described a brain system involving the medial prefrontal cortex, inferior frontal gyrus (IFG), superior temporal sulcus and temporal pole in ToM processing [7] [26]. In addition, the role of IFG in ToM circuits has been thought to block self-centered reasoning [8] [10] [27] [28]. YA sources were located at all these areas (Figure 2 and Figure 3) and while IFG located at BA 47 is common to both YA and NA analysis, IFG located at BA 45 is part of P 2 in YA case but not in NA case. Taken together, these results corroborate the hypothesis that socially driven neural circuits are important for establishing the veracity of YA.
In contrast with the above findings, NA arguments were self-interest driven, which is typical of arguments against any gun control in a variety of societies. Humans know others as outside speakers, visual images and recognized actors stored in our retrospective memory. Our own selves are identified by inner speech, somesthetic imagery, actions and episodic/autobiographic memories. The superior frontal and lingual gyrus have been reported [24] [29]- [32] as important structures associated with self-intended actions. The precuneus is a structure considered as a hub connecting the parietal and prefrontal regions [33] [34], and thereby plays an important role in various functions, including retrospective, episodic and autobiographic memories [7] and self reasoning [7] [10] [27]. Many of the sources discriminating NA from YA are located in the superior frontal, fusiform and lingual regions as well as in the precuneus, supporting the hypothesis that self-related circuits are important in establishing NA truth. The precuneus is also among the structures where YA discriminating sources were located; however, these sources were located in Brodmann area 7 and 19 in the YA and NA cases, respectively. Furthermore, in the rectal gyrus, which is one of the regions where the NA discriminating sources were located, the authors proposed that it is involved in evaluating the degree of NA truth. These results corroborated the hypothesis that self-interest driven neural circuits are important for establishing the veracity of NA.
Patterns P 2 and P 3 dramatically changed from truth to justification evaluation in NA case. P 2 and P 3 are associated to sources located predominantly at the left hemisphere during truth evaluation and bilaterally located in case of justification analysis. In this case, P 3 is also associated to sources located at more anterior places in comparison to truth evaluation. All patterns were similar during justification and influence evaluation for NA. As discussed above truth evaluation of NA implies logical inference whereas justification analysis implies to estimate expected benefits and risks. Because of this, neural circuits supporting each type of analysis to be different as observed here. In addition, NA were believed by more than 75% volunteers but 75% of then considered that they did not justify voting No. Therefore, it may be supposed that expected risks were considered greater than benefits. Because NAs were considered not justified they would not influence vote decision. Because risks were supposed greater than benefits NAs would not influence vote decision. This may explain because brain activity was similar during both J and I epochs.
Pattern P 3 did not change from truth to justification evaluation in YA case but P 2 was associated to sources located predominantly at the left hemisphere during truth evaluation and bilaterally located in case of justification analysis. In addition, pattern P 2 did not change from justification to influence evaluation in YA case but P 3 was associated with sources located at a small number of locations during influence than justification evaluation.
To be considered as justified, YAs have to be true and frequent. Around 50% of the volunteers considered them as true, but less than 30% of them considered that arguments would justify vote. These means that most of volunteers did not considered the appointed fact as frequent or associated with an important social damage. But these evaluations are also of empirical nature. To be true the probability of the proposed fact has to be greater than 0, and to be considered frequent it must be much greater than 0. So, the same circuits must be involved in both analyses. Because YAs were considered not justified they would not influence vote decision as observed here. This may explain why because brain activity may involve a smaller number of areas during I epoch in comparison to J epoch.

Conclusion
In conclusion, it may be said that our results show that different neural circuits support YA and NA analyses because these analyses enrolled neurons located at different cortical areas. YA analysis was associated with activation of cortical areas known to be involved in retrospective and episodic memory as well as with areas known to be involved in evaluating others' intention. In contrast, NA analysis was associated with cortical areas that are proposed to be involved with self-interest evaluation. Moreover, information about neural mechanism involved in voting decision-making obtained in this work helped to differentiate distinct ontological origins of YA and NA.