*Article* **Cerebral Arterial Asymmetries in the Neonate: Insight into the Pathogenesis of Stroke**

**Anica Jansen van Vuuren 1, Michael Saling 1,2,\*, Sheryle Rogerson 3, Peter Anderson 4,5, Jeanie Cheong 3,6 and Mark Solms <sup>7</sup>**


**Abstract:** Neonatal and adult strokes are more common in the left than in the right cerebral hemisphere in the middle cerebral arterial territory, and adult extracranial and intracranial vessels are systematically left-dominant. The aim of the research reported here was to determine whether the asymmetric vascular ground plan found in adults was present in healthy term neonates (n = 97). A new transcranial Doppler ultrasonography dual-view scanning protocol, with concurrent B-flow and pulsed wave imaging, acquired multivariate data on the neonatal middle cerebral arterial structure and function. This study documents for the first-time systematic asymmetries in the middle cerebral artery origin and distal trunk of healthy term neonates and identifies commensurately asymmetric hemodynamic vulnerabilities. A systematic leftward arterial dominance was found in the arterial caliber and cortically directed blood flow. The endothelial wall shear stress was also asymmetric across the midline and varied according to vessels' geometry. We conclude that the arterial structure and blood supply in the brain are laterally asymmetric in newborns. Unfavorable shearing forces, which are a by-product of the arterial asymmetries described here, might contribute to a greater risk of cerebrovascular pathology in the left hemisphere.

**Keywords:** middle cerebral artery; diameter; blood flow; asymmetry; stroke; shear stress; neonate

#### **1. Introduction**

Middle cerebral artery strokes occur more commonly in the left cerebral hemisphere [1,2]. In the mature brain, this leftward predilection has been attributed by some to selective recognition of the clinically obvious sequalae of left hemispheric events [3]. Neurovascular vulnerabilities that might explain a left hemispheric predilection for stroke have also been identified [4].

Adult studies report left-biased asymmetries in the structure and hemodynamics of extracranial and intracranial arteries, namely, the vertebral arteries [5,6], common and internal carotid arteries [7], and middle and anterior cerebral arteries [8]. These reports of larger arterial calibers, higher flow velocities, and higher blood flow volumes on the left are in keeping with the notion of a more resource intensive left hemisphere [9] and create left–right differences in the circulations of each arterial tree.

Hemodynamic processes, such as changes in blood pressure parameters, the speed of the pressure wave propagation, and resulting shearing forces on the arterial endothelium,

**Citation:** van Vuuren, A.J.; Saling, M.; Rogerson, S.; Anderson, P.; Cheong, J.; Solms, M. Cerebral Arterial Asymmetries in the Neonate: Insight into the Pathogenesis of Stroke. *Symmetry* **2022**, *14*, 456. https:// doi.org/10.3390/sym14030456

Academic Editor: Thierry Paillard

Received: 22 December 2021 Accepted: 24 January 2022 Published: 24 February 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

play important roles in the development of vascular disease [10]. The distribution of atherosclerosis in the vascular system is not uniform and plaque severity and composition also varies according to location [11]. Reports of higher left-than-right intima-media wall thickness [12] and plaque incidence, thickness, and instability [3] in the carotid arteries suggest a lateralized vulnerability to cerebrovascular disease in adults.

A left hemisphere predilection for cerebrovascular pathology, such as periventricular hemorrhage [13], neonatal stroke [14], and cerebral palsy [15], has also been reported in neonates. Approximately 70% to 80% of neonatal ischemic strokes occur in the middle cerebral arterial field, and they are left-sided in 53% to 75% of cases [16]. This begs a key question: is the ground plan of adult arterial asymmetries and corresponding vulnerability to pathology discernible in neonates? There is only one study, to our knowledge, that aimed to investigate the significance of left–right differences in the blood flow velocity to neonatal stroke, but only 20 normal control cases were reported, without data on arterial diameter, flow volume, or shear stress [17].

Ultrasonography of neonatal cerebral arteries is common in routine clinical practice and largely proceeds on the assumption of trans-midline symmetry. The resolution of existing methodologies has not been extended to detect the existence of structurofunctional asymmetries]. In previously used Doppler technologies, "bleeding", blooming artefact, and the influence of gain settings are recognized sources of error, particularly in relation to diametric measurement. This is problematic, since conclusions about regional cerebral blood flow cannot be drawn from velocity measurements [18], primarily because the volume flow (Q) in a vessel is related to the velocity (V) as well as the vessel's radius (R) according to the equation Q = VπR2. Similarly, the calculation of the wall shear stress requires a diametric measurement according to the equation τ = 4μ(V/πD3). B-flow imaging is a recently introduced non-Doppler technology which effectively bypasses these difficulties [19]. A dual-view imaging protocol using concurrent pulsed wave and B-flow Doppler transcranial ultrasonography addresses these shortcomings and paves the way for investigating the aims of the research reported here, namely, to investigate neonatal arterial asymmetry and corresponding cerebrovascular vulnerabilities.

We focused on the trunk of the middle cerebral artery as a major and accessible conduit to the lateral neocortical territory. We hypothesized that the diameter, hemodynamics, and shear stress are all inherently asymmetric in the direction of larger arterial calibers, higher blood flow volumes, and unfavorable shear stress on the left in the majority of healthy term neonates.

#### **2. Materials and Methods**

#### *2.1. Search Strategy and Selection Criteria*

Transcranial Doppler ultrasonography was performed on 106 healthy term neonates. Neonates with a gestational age greater than 37 weeks were recruited consecutively between March 2017 and November 2017 from the postnatal wards of the Royal Women's Hospital and Frances Perry House in Melbourne, Australia. A non-randomized participant sampling approach accompanied by comprehensive exclusion criteria (see below) was adopted. Six participants were excluded from the final analysis because aberrant middle cerebral arterial branching patterns precluded left–right comparisons of arterial geometry and hemodynamics. An additional three participants were excluded for poor image quality because of excessive neonatal movement, excessive hair, and or small cranial windows.

#### *2.2. Neonatal Exclusion Criteria*

Infants with significant perinatal complications were excluded (for example, postnatal resuscitation and/or admission to the neonatal intensive and special care nursery). Neonates with an intracranial pathology, substance exposure, or metabolic, genetic, and/or cardiovascular disorders were excluded. All infants enrolled in the study were healthy and had no dysmorphic features during the neonatal predischarge check.

#### *2.3. Maternal Exclusion Criteria*

Exclusion criteria included diagnoses of autoimmune disorders, pre-gestational diabetes mellitus, gestational diabetes, cardiac disease, drug and substance use, instances of suspected or detected fetal abnormality prior to delivery, chronic or persistent hypertension (>140/90), infections (including active genital herpes, syphilis, and HIV+), pre-eclampsia, and neurological and mental health conditions. Non-English-speaking parents were excluded from the study to ensure effective communication and understanding between the parent and investigators.

All scanning took place at the Royal Women's Hospital, Melbourne, Australia. Ethical approval was granted by the Royal Women's Hospital Human Research Ethics Committee and written informed consent was obtained from one or both parents.

#### *2.4. Procedure*

Transcranial ultrasonography and Doppler assessment took place at a postnatal age of 1 to 7 days. Scans did not reflect acute hemodynamic changes known to occur in the first 12 h of life [20]. Standard medical procedure was followed prior to the analysis. All infants underwent 10 min of supine rest on a clean cot in a standardized sound proofed ultrasound room with no auditory or visual distractions. The room had constant illumination and comfortable room temperature. Neonates were swaddled and fed prior to the scanning session. Parents were positioned at the head of the cot, behind the investigator so as not to distract the infant. If the neonate began to cry, the neonate was soothed before resuming the procedure.

Transcranial Doppler cerebrovascular imaging was performed using the portable General Electric EPIQ 9 ultrasound unit (GE Healthcare, Wauwatosa, WI, USA). A C3- 10-D convex probe (2–11 MHz) with an insonation angle close to 0◦ was used. Further settings included a small sample volume of 2 mm with a velocity wall filter of 80–100 Hz to eliminate noise caused by vessel wall movement.

Using a trans-temporal approach, the middle cerebral artery trunk was located by placing the transducer on the left temporal bone, below the zygomatic arch. Screening for previously undetected pathology and identification of the middle cerebral artery was performed with two-dimensional B mode gray-scale imaging and color flow imaging through the temporal window. B-flow imaging was activated, and the probe was moved so as to optimize the visualization of the origin of the middle cerebral artery trunk (approximately 2 mm from internal carotid artery terminus). At this distance, the vessel has a uniform diameter and required minimal angle correction. In any necessary instance, an angle of correction was performed if the angle of incidence was greater than 15◦ to ensure the transducer remained parallel to the vector of blood flow and accurate measures were obtained. Dual-view imaging was then initiated to replicate the image into two identical left and right images. The left image was selected, pulsed wave Doppler was activated, and several hemodynamic measurements were recorded at the arterial site. B-flow and pulsed wave frequencies used were 6.0 MHz and 4.2 MHz respectively. Three distinct pulsed wave spectral tracings containing three consecutive cardiac cycles were recorded. Peak systolic velocity (PSV), end-diastolic velocity (EDV), time averaged maximum velocity (TAMAX), time averaged mean velocity (TAMEAN), and heart rate measures were obtained. An on-site arterial diameter was taken from the corresponding right B-Flow image in the exact location hemodynamic measures were sourced.

The distal portion of the middle cerebral artery trunk (distal to the origins of the lenticulostriate arteries) approximately 2 mm from the middle cerebral artery bifurcation/trifurcation was located, and hemodynamic and diameter measures were repeated. The procedure was then repeated on the contralateral Mo and MDT sites. The sequence of data collection from the left and right middle cerebral arteries was randomized.

As a proof of concept for the new scanning protocol, we also imaged the very fine lenticulostriate branches of the middle cerebral trunk to a high degree of resolution. Lenticulostriate artery sampling in the study was sparse, largely because these vessels are difficult

to image and of a small caliber. The fact that lenticulostriate arteries were imaged to the point of supporting reliable measurement attests to the resolution of the innovations that were introduced to accomplish this.

The lenticulostriate arteries of the left and right cerebral hemisphere were approached by placing the same C3-10 transducer in the mid-sagittal plane of the anterior fontanelle. The transducer was fanned into the left cerebral hemisphere. Screening and identification of the lenticulostriate arteries was performed with two-dimensional B mode gray-scale imaging and color flow imaging. B-flow imaging was activated and two lenticulostriate arteries in each cerebral hemisphere were chosen for further scanning based on the clarity of the image and orientation of the vessel (that is, the two arteries on each side that were most oriented in the vertical plane). The probe was moved so as to optimize the visualization of one of the selected vessels. B-flow and pulsed wave Doppler was utilized in dual-view imaging to record structural and hemodynamic measures of the lenticulostriate artery. The procedure was then repeated for the second unilateral and two contralateral lenticulostriate arteries in a randomized order.

All images were stored on optical disc for off-line analysis using SYNAPSE (PACS) 64-bit imaging software. All hemodynamic measures were averaged across three homogenous consecutive cardiac cycles for each arterial site. Further investigation of arterial diameter was performed offline with RadiAnt DICOM viewer (64-bit) imaging software (version 4.2.1). The mean lumen diameter of each arterial site was determined by averaging three independent measurements taken at the same location as on-line analyses. Parameters were also averaged across the ipsilateral origin and distal trunk of middle cerebral artery (MCAMEAN). Assessment of inter-rater reliability was performed by SR, an experienced sonographer, on 10% of participants randomly selected from the sample throughout the data collection period. Cronbach's alpha showed a high internal consistency of 0.963.

At each site, the following hemodynamic indices were calculated with the following formulae. Mean velocity (*VMEAN*):

$$V\_{MEAN} = \frac{PSV + EDV}{2}$$

Resistive index (*RI*) computed according to the method of Pourcelot (1982):

$$RI = \frac{PSV - EDV}{PSV}$$

Pulsatility index (*PI*) computed according to the method of Gosling and King (1988):

$$PI = \frac{PSV - EDV}{V\_{MEAN}}$$

Shear stress (*τ*; dyne/cm2):

$$\tau = 4 \cdot \mu \cdot \frac{V}{\pi \cdot D^3}$$

where *V* equals the flow velocity, μ equals the viscosity of flow, and *D* equals the arterial diameter. No data was available concerning blood viscosity in the neonates, so an average neonatal hematocrit-adjusted (0–45) blood viscosity, adjusted at high shear rates of 4.22 mPa.s, was assumed [21], as there is no reason to suspect intraindividual viscosity differences or systematic differences between left- and right-dominant neonates.

Volume flow (*Q*):

$$Q = PSV \times \left( D^2 \left( \frac{\pi}{4} \right) \right)^2$$

Peak systolic velocity was used as a variable in the calculation of volume flow because it is sensitive to left–right differences in the neonate [22], is mediated by arterial structure [23], and reflects cerebral blood flow [24], the definition of which is the primary aim of this work. Average measures (such as TAMEAN) inevitably conflate peak systolic velocity

with end-diastolic velocity. While this might be useful in particular clinical applications, end-diastolic velocities show less left–right differentiation [25].

Arterial diameter was used as a grouping variable for the sample. Interhemispheric diameter dominance was expressed in the form of a left–right laterality index and calculated with the formula:

$$Lateral \, index = \frac{L - R}{L + R}$$

where *R* equals the right arterial measure and *L* the left arterial measure. A positive value indicated left arterial dominance, whereas a negative value indicated right arterial dominance. A score of 0 represents the absence of a structural dominance. A LI was calculated for each arterial site as well as the cerebral artery average between the middle cerebral origin and distal trunk (MCAMEAN).

#### *2.5. Statistical Analyses*

Data were analyzed using IBM SPSS Statistics (version 23) software. Each hemodynamic measure of the middle cerebral arteries was analyzed using a mixed-design ANOVA. Neonates with no structural arterial dominance were removed from the analysis. For each analysis, the within-subjects factor was the respective arterial parameter (of the left and right paired arteries) and the between-subjects factor was the structural dominance (left-dominant or right-dominant). One-tailed paired t-tests compared lateral differences in geometric groups in instances of significant interactions. One-tailed independent t-tests also compared sex differences in participant demographics and hemodynamics parameters at each site of measurement. Tests of normality and homoscedasticity (namely Levene's test of equality of variance and Shapiro–Wilk tests) were run on each dataset. If the assumption of normality was not upheld, a non-parametric Mann–Whitney U test was run instead.

For brevity, results for TAMAX and TAMEAN are omitted as they are collinear with PSV (*r* > 0.90) and VMEAN (*r* > 0.90), respectively, as described below, but might not be as precise as PSV in defining lateral difference (see above).

Internal consistency was calculated using Cronbach's alpha in 10% of cases. Cohen's rule of thumb for effect size interpretations was used for between-group comparisons: *d* = 0.10 (small effect), *d* = 0.30 (medium effect), and *d* = 0.50 (large effect). The significance of the analyses was determined with a 95% confidence level at *p* < 0.05.

#### **3. Results**

The geometric and hemodynamic properties of the middle cerebral artery origin and termination of its trunk were recorded in 97 healthy full-term neonates. The final sample included 59 males and 38 females born via normal vaginal delivery or caesarean section (Table 1). The gestational age at birth of the sample ranged from 36 to 41 weeks, and birth weights ranged from 2200 g to 4930 g. The postnatal age at the time of scanning was 12 to 174 h (*M* = 47.71 h; *SD* = 28.58; *Median* = 41 h; *Range* = 162 h). Mean Apgar scores were 8.30 at 1 min (*SD* = 1.38) and 8.92 at 5 min (*SD* = 0.32). The sample spent an average of 68 h in hospital. There were no significant sex differences in the birth weight, postnatal scanning age, or gestational age.

As an example of dual views of B-mode and pulsed wave Doppler ultrasound imaging, Figure 1 illustrates the diameter and hemodynamic variability of the left and right middle cerebral arteries. As a proof of concept, we also imaged the very fine lenticulostriate branches of the middle cerebral trunk to a high degree of resolution. Example images are included in Figure 1. Demographic information on the neonates according to their averaged middle cerebral geometric asymmetry (left-dominant or right-dominant) is presented in Table 1.


**Table 1.** Neonatal characteristics as a function of geometric arterial asymmetry.

Note: AS1min = Apgar score at 1 min; AS5min = Apgar score at 5 min; <sup>a</sup> = number according to geometric dominance averaged across middle cerebral origin and distal trunk.

**Figure 1.** An example of dual-view B-flow and pulsed wave imaging in a left-dominant infant. The brown-scale arterial images and the blue-scale cardiac cycles for measurement of velocities are shown.

Panels (**A**): origin of the left and right middle cerebral arteries. The diameter on the left is 2.9 mm and the diameter on the right is 2.1 mm; the difference is visible on inspection of the brown-scale images. Peak systolic velocity on the left (PS in the quantitative panel) is 47.6 cm/s and 40.3 cm/s on the right. The end-diastolic velocity (ED in the quantitative panel) is 17.7 cm/s on the left and 11.2 cm/s on the right. Panels (**B**): distal segment of the trunk of the left and right middle cerebral arteries. The diameter on the left is 3.2 mm and 2.0 mm on the right, and the difference is again visible on inspection of the brown-scale images. Peak systolic velocity (PS) is 50 cm/s on the left and 40.3 cm/s on the right. End-diastolic velocity (ED) is 19.3 cm/s on the left and 12.8 cm/s on the right. Panels (**C**): the lenticulostriate arteries are shown largely as a proof of the concept that very small caliber arteries in the neonatal brain can be visualized and that structurofunctional measurements can be obtained. The arteries selected for measurement can be identified by the white dotted lines in the brown-scale images.

#### *3.1. Sex Differences*

No significant sex differences were found in the arterial diameter; peak systolic, end-diastolic, and mean velocity; and resistance or pulsatility indices. A significant sex difference was found in the blood flow volume in the left middle cerebral origin. Overall, males had higher left-sided blood flow volumes (*M* = 210.04 mL/min; *SD* = 75.05) than females (*M* = 189.87 mm; *SD* = 80.50) at this arterial site (*p* = 0.041). Shearing forces at each corresponding arterial site were comparable between males and females apart from the shear stress in the distal trunk of the right middle cerebral artery. Females had a higher right-sided wall shear stress (*M* = 595.99 dyne/cm2; *SD* = 208.90) than males (*M* = 504.71 dyne/cm2; *SD* = 137.61) at the distal trunk (*p* = 0.034).

#### *3.2. Structural Differences*

Left–right asymmetries in the arterial diameter were found at each arterial site (*p* < 0.001; Table 2). Of the 97 participating neonates, a left geometric dominance was exhibited in 52 (54%) at the middle cerebral origin and 60 (62%) at the middle cerebral distal trunk. When averaged across the arteries, with no consideration of individual dominance, a significant leftward structural difference was evident only at the middle cerebral distal trunk (t(96) = 1.989, *p* = 0.050, *d* = 0.239). A small proportion of participants showed no left–right differences in the arterial diameter at the origin (8%) and distal trunk (4%). Laterality indices of structure at the middle cerebral artery proximal segment were associated with asymmetries at the distal segment (*r* = 0.741; *p* < 0.001).

Analyses described in this paper have not been undertaken in previous work. Rather, left–right comparisons classically are made on the basis of average values across the entire sample and with measurements taken at a single site, namely the origin of the middle cerebral artery. The findings reported in the "Averaged" column of Table 2 show that this approach hides or reduces the probability of the systematic individual lateral dominance reported here.

#### *3.3. Structurofunctional Differences*

Considerable geometric and hemodynamic asymmetries existed in the origin and distal trunk of the middle cerebral artery. In participants with a leftward dominance in the arterial geometry, peak blood flow velocities were higher on the left side at both sites of the middle cerebral artery. A leftward bias in the average flow velocity was found at the origin, and higher blood flow volumes were also found in the larger left origin and distal trunk of this group. No lateral differences in the end-diastolic velocity were found at either site.

Across both middle cerebral arterial sites, no lateral differences in the peak systolic, end-diastolic of average blood flow velocity were found in neonates with larger arteries on the right side (Tables 3 and 4; Figure 2). Converse to neonates with a leftward dominance in the geometry, a right-sided asymmetry in the overall blood flow volume was found at the origin and distal trunk of neonates with larger arteries in the right hemisphere.


**Table 2.** Intra-individual left–right diametric differences as a function of inter-individual differences in the direction of arterial asymmetry.

Note: *p* < 0.05; MCAO = middle cerebral artery origin; MCADT = middle cerebral artery distal trunk; MCAMEAN = middle cerebral artery averaged across origin and distal trunk measures; <sup>a</sup> = averaged across the sample with no consideration of individual differences in arterial asymmetry.

**Table 3.** Comparisons of hemodynamic parameters between left and arterial sites according to geometric dominance.


**Table 3.** *Cont.*


Note: *p* < 0.05; MCAO = middle cerebral artery origin; MCADT = middle cerebral artery distal trunk; PSV = peak systolic velocity; EDV = end diastolic velocity; VMEAN = mean velocity; RI = resistance index; PI = pulsatility index; Q = blood flow volume; WSSSYS = systolic wall shear stress; WSSDIAS = diastolic wall shear stress; \* = significant *p* values.




**Table 4.** *Cont.*

Note. *p* < 0.05; PSV = peak systolic velocity; EDV = end diastolic velocity; VMEAN = mean velocity; RI = resistance index; PI = pulsatility index; Q = blood flow volume; WSSSYS = systolic wall shear stress; WSSDIAS = diastolic wall shear stress; \* significant *p* values

**Figure 2.** Interactions between haemodynamic and arterial wall shear stress variables and left versus right middle cerebral arteries at the origin (**A**–**D**) and distal trunk (**E**–**H**) in left and right dominant neonates. The haemodynamic variables are peak systolic volume (**A**,**E**) and blood flow volume (**B**,**F**). The shear stress variables are systolic wall shear stress (**C**,**G**) and diastolic wall shear stress (**D**,**H**). Abbreviations: LMCA = Left middle cerebral artery; RMCA = Right middle cerebral artery; PSV = Peak systolic velocity; Q = blood flow volume; WSSSYS = systolic wall shear stress; WSSDIAS = diastolic wall shear stress. Error bars show the 95% confidence interval.

The influence of the neonatal arterial geometry on the hemodynamics of these two sites varied, in that the effect of a structural dominance was more pervasive at the origin across most blood flow velocity and flow volume measures (Tables 3 and 4; Figure 2). More specifically, interactions were found in the peak systolic velocity, average velocity, and blood flow volume. Structural dominance of the distal middle cerebral trunk did not significantly influence the arterial velocity (peak systolic, end-diastolic, and mean velocities), but a significant influence of the geometry was seen in the blood flow volume.

The resistance to the blood flow caused by the microvascular bed distal to the site of measurement did not significantly interact with the arterial geometry (Table 4). However, the main effect on the arterial resistance was found at the middle cerebral distal trunk, where the resistance distal to the middle cerebral artery trunk terminus was higher across both cerebral hemispheres in neonates with a rightward geometric dominance (*p* = 0.031). This main effect was also reflected in pulsatility indices.

No lateral differences in the arterial resistance were noted at the middle cerebral artery origin. A lateral difference in the arterial resistance was evident at the distal trunk, where neonates with a leftward structural dominance had a higher resistance and pulsatility index in the left cerebral hemisphere than they had in the right (Table 3). No left–right differences were found in those with a rightward arterial dominance.

#### *3.4. Shearing Stress Differences*

The neonatal arterial geometry differentially influenced peak systolic and end diastolic shearing forces at both arterial sites (Table 4; Figure 2). In participants with a leftward structural dominance, the peak systolic and end diastolic shear stresses were significantly higher on the right-side than on the left, and the converse was seen in right-dominant neonates.

#### **4. Discussion**

In adults, left–right asymmetries are normal attributes of cerebral perfusion, akin to well-established asymmetries in brain morphology [26,27]. Cerebral arterial diameters and blood flows have been investigated in neonates for a variety of largely clinical ends. Studies of diameters are restricted to autopsy series [28,29]. Blood flow velocity is commonly measured in vivo for routine clinical purposes [30–35].

To our knowledge, this is the first intentional investigation of structurofunctional neonatal cerebral arterial asymmetries in healthy term neonates at rest. Differences in diameter were found at each arterial site of interest, and the corresponding hemodynamics were biased towards larger arterial calibers. Leftward hemodynamic biases were found in neonates with larger arteries in the left cerebral hemisphere (left-dominant), while rightward hemodynamic biases were found in neonates with larger arteries in the right cerebral hemisphere (right-dominant). Very few neonates (<8%) showed an absence of lateral differences in arterial diameter.

The pattern of asymmetry in middle cerebral Doppler waveforms differed between left- and right-dominant groups. Left-dominant neonates were typified by impressive differences in left–right peak systolic velocities that disappeared at the end systole. This peak systolic effect was absent bilaterally in neonates with larger arterial diameters on the right.

Although pulsatility and resistance indices are frequently used in clinical studies, the interpretation of these variables is dependent on several factors such as vascular resistance, arterial compliance, and the driving force of the arterial pulse wave [36]. Structural dominance did not play a role in resistance and pulsatility differences. Arterial pulsatility was not laterally biased in right-dominant neonates, but in left-dominant neonates pulsatility was left biased in the distal trunk of the middle cerebral artery. If one were to apply a traditional interpretation [37] to these findings, the degree of resistance in the cortical microvascular bed distal to the middle cerebral artery would be predicted to be higher in the left hemisphere of most neonates. Higher indices in the left middle cerebral artery would, in turn, indicate a decreased end-diastolic velocity, rendering the left hemisphere more prone to disorders such as stroke or venous infarcts, and left-biased resistance and pulsatility asymmetries in neonates have been documented previously [38].

The arterial endothelial wall shear stress exerts a key influence on the genesis of vascular pathology [39], in that a high shear stress has a protective effect on the endothelium [40]. The pathogenesis of the higher left-than-right incidence of cerebrovascular pathology in adults [41,42] and neonates [43] has been elusive, but clarification might be gained from the overall blood flow and wall shear stress asymmetries reported here.

The present findings show that a systematic leftward arterial bias in wall shear stress is detectible in healthy term neonates. Shearing asymmetries systematically disadvantaged the left hemispheric endothelium with lower left-than-right peak systolic and end-diastolic endothelial shearing forces in neonates with larger left-sided arteries. The converse was seen in right-dominant neonates. This asymmetry therefore increased the neurovascular vulnerability in the left cerebral hemisphere of most healthy term neonates.

The findings of this study are consistent with the hypothesis that the wall shear stress varies according to geometric and behavioral lateralization in the neonatal cerebral arterial trunk. This adds to literature [44] demonstrating that the wall shear stress varies with location across the cardiovascular system. These findings therefore bring Murray's law of constant shearing forces throughout the arterial system [45] into question.

The ontogenesis of atherosclerosis begins very early in life. While incipient atherosclerotic changes are minor in most cases, the process can be accelerated in the presence of a variety of conditions [46]. Menshawi and colleagues [47] postulated that individuals born with an unfavorable arterial geometry are more susceptible to the atherosclerotic effects of traditional vascular risk factors. Although left-lateralized lesions are not inevitable, the predisposing effects of "atherosclerosis-enabling" anatomy reported here might provide the framework for a greater left-than-right incidence of cerebrovascular pathology.

If the present findings are stable across the lifespan and are also consistently discernible in adults, extended exposure to a lateralized arterial vulnerability might also shed light on the ontogenesis of leftward biases in the carotid intima-media wall thickness [48]; plaque incidence, thickness, and instability, as well as large-vessel ischemic events in adults [3].

There are documented associations between handedness and the left arterial intimal wall thickness of the carotids [49], as well as left-handedness and a lower risk of sudden death from brain infarction (typically associated with left-hemispheric stroke [50]. In an adult study, some of the present authors showed that the arterial length, diameter, resistance to blood flow, velocity, and volume flow rate are asymmetric and are intimately related to hand preference and proficiency, raising the possibility that these structurofunctional asymmetries arose in adaptation to greater metabolic demands in the dominant hemisphere in anticipation of the emergence of lateralized cognitive and behavioral functions [51].

Our data show that the asymmetric vascular ground plan found in adults is present in neonates. Ultimately, routine investigations of the neonatal brain should proceed on the expectation that asymmetries in the middle cerebral arteries are a normal attribute of the lateral cortical supply. Ironically, the lateralized neurovascular framework within which language develops might also contain the seeds of its most significant cerebrovascular threat.

**Author Contributions:** Conceptualization, A.J.v.V., M.S. (Michael Saling) and M.S. (Mark Solms); methodology, A.J.v.V. and S.R.; software, A.J.v.V. and S.R.; validation, S.R.; formal analysis, A.J.v.V.; investigation, A.J.v.V.; resources, S.R. and J.C.; data curation, A.J.v.V.; writing—original draft preparation, A.J.v.V. and M.S. (Michael Saling); writing—review and editing, A.J.v.V., M.S. (Michael Saling), P.A., S.R., M.S. (Mark Solms) and J.C.; visualization, A.J.v.V.; supervision, M.S. (Michael Saling); project administration, A.J.v.V.; funding acquisition, M.S. (Mark Solms). All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** Royal Women's Hospital Human Research Ethics Committee Project 16/18—Cerebral arterial asymmetries in the neonates.

**Informed Consent Statement:** Informed consent was obtained from the parents or legal guardians of all neonates involved in the study.

**Data Availability Statement:** The data presented in this study are available on request from the corresponding author. The data are not publicly available due to privacy reasons.

**Acknowledgments:** We are indebted the Royal Women's Hospital and Frances Perry House for providing the platform for us to conduct our study. We would also like to thank Julie Archbold for her help in calibrating the Doppler ultrasound machine. The financial assistance of the University of Melbourne is acknowledged.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


### *Review* **The Neurological Asymmetry of Self-Face Recognition**

**Aleksandra Janowska, Brianna Balugas, Matthew Pardillo, Victoria Mistretta, Katherine Chavarria, Janet Brenya, Taylor Shelansky, Vanessa Martinez, Kitty Pagano, Nathira Ahmad, Samantha Zorns, Abigail Straus, Sarah Sierra and Julian Paul Keenan \***

> The Cognitive Neuroimaging Laboratory, Montclair State University, 207 Science Hall, Montclair, NJ 07043, USA; janowskaa1@montclair.edu (A.J.); balugasb1@mail.montclair.edu (B.B.); pardillom1@montclair.edu (M.P.); 220731@pcti.mobi (V.M.); chavarriak2@mail.montclair.edu (K.C.); brenyaj1@mail.montclair.edu (J.B.); shelanskeyt1@montclair.edu (T.S.); martinezv8@montclair.edu (V.M.); paganok1@mail.montclair.edu (K.P.); ahmadn3@mail.montclair.edu (N.A.); zornss1@montclair.edu (S.Z.); astra20030@sufsdny.org (A.S.); sierras5@montclair.edu (S.S.)

**\*** Correspondence: keenanj@montclair.edu

**Abstract:** While the desire to uncover the neural correlates of consciousness has taken numerous directions, self-face recognition has been a constant in attempts to isolate aspects of self-awareness. The neuroimaging revolution of the 1990s brought about systematic attempts to isolate the underlying neural basis of self-face recognition. These studies, including some of the first fMRI (functional magnetic resonance imaging) examinations, revealed a right-hemisphere bias for self-face recognition in a diverse set of regions including the insula, the dorsal frontal lobe, the temporal parietal junction, and the medial temporal cortex. In this systematic review, we provide confirmation of these data (which are correlational) which were provided by TMS (transcranial magnetic stimulation) and patients in which direct inhibition or ablation of right-hemisphere regions leads to a disruption or absence of self-face recognition. These data are consistent with a number of theories including a righthemisphere dominance for self-awareness and/or a right-hemisphere specialization for identifying significant social relationships, including to oneself.

**Keywords:** symmetry; self-face recognition; right hemisphere; self-awareness

#### **1. Introduction**

The evolution of animal nervous system symmetry is complex, with many resulting variants [1–5]. Allowing for numerous phenotypic advantages, including those at both an individual and social/interactive level [6–9], the nervous systems of bilateral organisms have exploited the benefits of a lateralized nervous system for hundreds of millions of years [2,10] (but see [11]).

The human brain is no exception [2,12]. The first impression of the human brain was noted as far back as the Ancient Greeks as two distinct hemispheres. Except in rare cases of severe abnormal development, a human at any stage post-second trimester will anatomically have two distinctly visible hemispheres. These anatomical differences have given rise to functional differences, scientifically noted by Broca, Wernicke, and others in the late 1800s and early 1900s [13]. While language remains the most well-known of human brain lateralization, many other functions appear distinctly prominent in one hemisphere [14–18].

That being said, research concerning left and right hemisphere differences (LH/RH) in the brain appears to trend from 'too simplistic' to 'explains everything'. While Roger Sperry's Nobel Prize in 1981 seemed to cement the legitimacy of exploring hemispheric differences [19], the popular press has run with mythical notions such as people being 'right- or left-brained'. The more measured approach is understanding both the ultimate and proximate reasons for asymmetries.

**Citation:** Janowska, A.; Balugas, B.; Pardillo, M.; Mistretta, V.; Chavarria, K.; Brenya, J.; Shelansky, T.; Martinez, V.; Pagano, K.; Ahmad, N.; et al. The Neurological Asymmetry of Self-Face Recognition. *Symmetry* **2021**, *13*, 1135. https://doi.org/10.3390/ sym13071135

Academic Editors: Sergei D. Odintsov and Chiara Spironelli

Received: 1 April 2021 Accepted: 21 June 2021 Published: 25 June 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

For example, the motor asymmetries observed in humans and other primates [12,20–22] have led to numerous theories the most useful of which include both the mechanisms and the underlying cost/benefit analyses. An example of this is that human cradling (mother/infant) is performed employing the left arm the majority of the time [8,23–25]. Such a bias is explained by socio-affective communication being RH dominant, which clearly taps into evolutionary explanations (i.e., facilitating social bonding) while also explaining functional brain hemispheric differences. That is, these data help to resolve why certain aspects of emotional communication may be RH dominant and how righthandedness itself may have evolved.

Here we describe the evidence that self-face recognition (SFR) is RH dominant and speculate that it is related to the underlying construct of self-awareness (SA; [26]), and we provide a review of the literature. It is recognized that while the evidence for RH dominance is robust [27], thus far the data are suggestive in terms of a link between SFR and SA [28–30]. In terms of evolution, we know even less. For example, while initial evidence indicates hemispheric differences in terms of SFR in chimpanzees [31,32], other animals (particularly the magpie, who may or may not have SFR: [33,34] may have an entirely different underlying neural structure that supports SFR and, potentially, SA. We therefore are unable to know at this point whether SFR has evolved independently or in a more homologous manner [35].

The cone of uncertainty widens in terms of SFR as we move phenotypically and phylogenetically further from *Homo sapiens*. In humans, we propose SFR as the benchmark against which all measures of self-awareness should be tested. We suggest, in fact, that despite numerous challenges, SFR in the great apes indicates SA, and we predict that in the near future, homologous neural underpinnings in the RH will be discovered that sustain SFR. We equally suggest that as we move to cetaceans, corvids, elephants, etc., SFR becomes more unstable, and both the underlying structures and the underlying indices of SFR become much less clear. Our purpose here is to highlight the main research that underlies these claims. While not a thorough review, this summary is intended to provide a clear and concise overview of SFR.

#### **2. The History of Self-Face Recognition: Measuring Self-Awareness**

Questions concerning self-awareness have been posed by almost all humans, including scientists, for millennia. From the Ancient Greek scholars, all the way through Gallup's mirror self-recognition tests in the 1970s, to today's modern brain imaging techniques, self-awareness has always been an intriguing topic to study and investigate.

Greek philosophers, including Socrates (b. 470 B.C.E.), believed that introspection was necessary for humans to be truly cognizant and pure. Plato (b. ca. 428 B.C.E.) took this concept of self even further and stated that introspection was a human obligation and that knowledge of "good" and "self" were needed in order to be honorable and principled. Importantly, Aristotle (b. 384 B.C.E.) took a comparative approach to look at differences between self-awareness in humans and nonhuman animals through studying cognitive intelligence. He concluded that both humans and animals had basic functions, such as sight, smell, taste, etc., but *pure* intellect was only found in humans, which made a large distinction between humans and animals. Additionally, Aristotle was one of the first to attempt to create a relationship between the self, soul, and body [36].

Most famously, the French mathematician René Descartes (b. 1596) took the study of consciousness further, as he is often considered among the first neuroscientists that attempted to localize the self. Many of his ideas are still commonly spoken of today, such as "*Cogito, ergo sum*" ("I think; therefore, I am"), which speculated that the self can exist because it can think of its own existence. Outside of just defining the self, Descartes attempted to actually locate the self in the brain. Although his determination of self in the brain as being located in the pineal gland (due to its centralized position in the brain) was ultimately wrong, his comparative look at humans and animals had a lasting influence [37–40]. Unlike Aristotle, Descartes believed that animals are intelligent but do

not have a soul or self. He found that animals do not use language, behave on impulses, and are not adaptable, so they cannot have a self.

The Ancient Greek philosophers and Descartes laid the groundwork for formalizing scientific investigations of the self under the umbrella of psychology. Most famously, Sigmund Freud explained the deep, buried unconscious mind by using the self while simultaneously explaining the self [36]. Carl Jung believed that there were many common selves that people shared to some degree. Jean Piaget believed that children refine their self through assimilation and childhood experiences, which, in turn, play a key role in growing into adulthood. Many philosophers (including Locke, Sartre, Hegel, and Hume) and psychologists (e.g., Seligman, Beck, and Kohler) examined relationships between self-awareness and cognition.

It is noted that many individuals were addressing self/other distinctions as other sciences came into their own. In the 18th century, Carolus Linnaeus, a Swedish botanist, created a binomial classification system for living organisms. While it is known that many of his taxonomic names are still used today, such as kingdom, order, species, etc., we note here that one important classification was grouping humans along with other animals such as monkeys, apes, and bats as primates. Linnaeus remembered a Latin inscription, "*Nosce te ipsum*" ("Know thyself"), translated from the Greek above the Temple of Apollo at Delphi, from which he assumed that the distinction between humans and other primates is the capability for self-recognition and self-knowledge. Therefore, he categorized humans into *Homo sapiens*, wise men, since he believed that self-awareness was the highest form of uniquely human intelligence [41,42].

Up until this time, the self was quite an abstract concept with little concrete evidence, but that slowly changed when the mirror was seen as a tool to be used to measure cognitive abilities and self-awareness. Grant, in 1828, was the first person to our knowledge to use a mirror for a self-recognition study. The study found that monkeys, in general, had a surprised reaction when looking at mirror glass, but orangutans, in particular, had no emotional response to looking at the glass. The exact reaction of the orangutans was not recorded, just the lack of reaction, which was unfortunate, but enough of a reaction to record considering the monkeys' strong reaction to the mirror. Soon after, Charles Darwin was one of the first to suggest using mirror recognition as a measure of higher cognitive abilities. His first recorded mirror test in 1840 examined the behavior of orangutans presented with a mirror, in which he recorded, and later published in 1876, that the orangutans would look at the mirror as if they were seeing another animal [43]. This led to Darwin's conclusion that self-recognition was not an ability of nonhuman animals. Additionally, Darwin studied his 10 children as they grew up, starting in 1839, and did mirror tests with them. His conclusions were that self-awareness and self-knowing were tied to the ability to self-recognize [44].

In 1878, Maximillian Schmidt reported similar findings to Grant and Darwin (that orangutans did not have a self-reaction to a mirror but seemed to understand the reflective properties of the mirror). Schmidt noted that the orangutan was able to identify a human reflection in the mirror of someone standing nearby [45]. These findings were the norm in many studies performed around this time with nonhuman mirror tests. Another study performed by J. von Fischer in 1876 observed monkeys and baboons in front of a mirror and, once again, a negative mirror self-recognition result was reported. It is noted that in these studies, rigorous methods to determine SFR were not employed. The general sense is that an organism was placed in front of a mirror and behavior was observed, such as attacking the mirror or apathy. The main conclusion was that the various monkeys tested by von Fischer reacted to the mirror as if it was another, novel monkey.

Not too long after, in 1889, Wilhelm Preyer, a German researcher, was able to define a definite sign of self through the use of mirrors. He studied how using only language would be an inadequate use to describe the self in children due to the lack of vocabulary, not the lack of understanding of the "I", or ego. He created developmental timelines using mirror recognition, language, and other time measures, such as own-shadow recognition, to pinpoint a child's timeline of self-recognition. Through this, Preyer was able to confirm the use of mirrors for self-recognition tests due to his orderly, thorough reports. Although Preyer did not specifically work with apes, he did record nonhuman animal reactions compared to the human reactions [46]. Unfortunately, most mirror self-recognition researchers did not communicate together at the time, so many went unnoticed, causing a dip in the study of mirror tests for self-recognition.

Outside the occasional mirror test on monkeys, orangutans, and chimpanzees, most mirror self-recognition slowed down, until 1929. Robert and Ada Yerkes found results suggesting no self-interest in mirrors by the nonhuman primates, noting that the animals were seeing 'another animal' in the mirror [47]. Following this, mirror tests were performed rarely and without any true lineage of experiments to follow. In 1940, C. W. Huntley also performed a little-known experiment that recorded human participants having a large emotional reaction to the realization that the recorded voice played back, hand pictures, and handwriting were indeed their own [48]. Arnold Gesell, a Yale child developmentalist, studied similar theories in the early to mid-20th century as Piaget and Preyer and studied many self-indicators of how a child's timeline develops these indicators [49]. Unfortunately, lack of expansion on these previous theories with mirror tests led Gesell's reports to fall through the cracks. In the 1940s, Jacques Lacan suggested that the formation of the self and mirror recognition were correlated, but due to a lack of mirror experimentation, his research was overlooked. In 1954, in a paper for the journal *Human Biology*, a photograph of a chimpanzee named Vicki using a mirror to guide pliers over her teeth was taken, but the authors, Keith and Catherine Hayes, did not discuss its relevance (that there was evidence that a *nonhuman had self-recognition*) and it went unrecognized [50].

#### **3. Gallup's Mirror Self-Recognition Test**

The lull of interest in mirror tests finally changed abruptly in 1970 when Gallup published research of a nonhuman-animal positive mirror self-recognition test. Gallup's mirror and mark test combined many previous ideas about self-recognition, and in the article, he commented that there is likely a connection between self-recognition and selfknowing. By creating a test that measured a real physical trait along with a well-thought-out process that eliminated random chance, Gallup made a solid argument for self-recognition in chimpanzees. However, he and others immediately picked up the notion that selfrecognition in a mirror may, in fact, be evidence that humans are not the only self-aware organism on this planet.

The first test consisted of a 10-day period of a mirror placed in the testing site with the chimpanzees to allow them to acclimate with the mirror. This created a baseline behavior of the chimpanzees, which was typical mirror behavior, as seen with previous tests where there was no significant reaction to the mirror. Then, the chimpanzees were placed under anesthesia and a small mark that was dry and odorless was placed above the eyebrow where the chimpanzees could not see directly. After waking up from anesthesia, the chimpanzees' behavior was observed without a mirror. It was found that the chimpanzees did not react to their new mark (in the absence of a mirror), no smelling or touching occurred, and therefore, it was concluded that they were unaware of the mark. After this short baseline period, mirrors were placed in the chimpanzee test site again and their reactions were observed. Upon seeing their reflections in the mirror, the chimpanzees would touch and smell the mark and investigate their hands after touching the mark. This indicated that the animals recognized that the mark had not been there previously and that they had to use a mirror to find the mark. This test itself was a breakthrough, but Gallup conducted two more tests to confirm this conclusion.

The second test was almost identical to the first, with the exception of the initial 10-day pre-mark mirror exposure. Gallup hypothesized that without the previous mirror exposure, the chimpanzees would not react to the mark as being odd on their face, which proved to be true. The chimpanzees were given the mark and indeed did not react to it in any significant manner, leading Gallup to conclude that the first group achieved mirror selfrecognition (MSR). The third test Gallup conducted was identical to the first, but instead of chimpanzees, he used monkeys. He observed their reactions pre-mark and post-mark with a mirror, and the monkeys gave no sign of MSR [51].

The three tests put together create a clear, concise result that chimpanzees are able to recognize themselves in the mirror. Gallup published his findings in 1970 not only discussing the test and the breakthrough results he found, but also contemplating the higher consciousness of nonhuman animals. Before then, it was assumed that humans have the highest form of intelligence and cognition and that animals have some intelligence but not necessarily a "soul" or self. With these findings, Gallup opened up the discussion of nonhuman animals possibly having a higher-order cognitive process and internal world. With this mirror and mark test, Gallup also sparked a new method of research with the ability to test consciousness hypotheses.

Since Gallup's initial findings, the literature has swayed from conservative (only humans, chimpanzees, and orangutans have self-face recognition) to today with more liberal interpretations that include the addition of animals such as elephants, dolphins, and magpies [52].

#### **4. SFR in Animals**

The advantages of Gallup's test are many. It does not require language, which means it can be used in pre-/non-linguistic humans and nonhumans. It requires no special equipment; the equipment needed is portable and can be used in any environment. It requires little training to conduct, it is inexpensive, and it does not take months to perform. The test is noninvasive and generally culture-fair, though there are cultures where mirror exposure may be limited [42].

Gallup's test involves a number of aspects that present difficulties in testing both human and nonhuman populations. The test relies on vision and memory, as well as intact motor systems, and alterations to these can influence results. The test requires the individual to inspect the mark to 'pass', which in the original form meant having limbs/arms capable of touching the mark. The test requires some aspect of 'caring' that one's image is altered to the point of touching the mark. This last point is a major issue when testing special groups of humans, such as those with autism, who may know that there is a mark but not feel compelled to inspect the mark [53,54]. While there are other issues (e.g., attentional processing, what qualifies as a successful touch, and the role of training vs. spontaneous SFR), it should be apparent that direct comparison across populations is difficult if not impossible.

For example, dolphins and other marine mammals are notorious for being a 'challenging' population to test as mark-directed responses are near impossible to conduct [55–57]. Self-recognition has been claimed not via mark-directed touching but via inspection of the mark in a mirror and self-directed behavior conducted by use of the mirror. Horses have been tested via 'scraping' behavior of a mark on the cheek [58]. Elephants are considered to pass if they make a trunk-directed response [59–61].

It is possible that tests for SFR are most opaque in terms of the avians [62–66]. While an early report indicated magpies passing a modified version of Gallup's test [64], this finding has not been consistent [33]. B.F. Skinner was one of the first to perform SFR avian studies when he and his colleagues conditioned pigeons to peck at a spot [67]; see [65,66], which caused others to emphasize that SFR must be spontaneous and not 'taught' [68]. Other attempts to modify the test have included making the test more phenotypically relevant by using olfaction rather than vision [69]. While this is but a brief summary, we remain skeptical about SFR in non-great apes (with the exception of dolphins) and conclude that more evidence is needed before we consider organisms outside of apes and dolphins to have SFR abilities [52].

#### **5. Nonhumans and Brain Symmetry**

Before presenting evidence of a lateralized system for SFR, it is worth examining the sparse evidence of the neural correlates of SFR in nonhumans. Bill Hopkins and his team examined mirror self-recognizers (MSR+) vs. non-recognizers (MSR−). Employing diffusion tensor imaging, a lateralized MSR+ vs. MSR− difference was found in the superior longitudinal fasciculus (including frontal lobe areas). The authors indicated that SFR was associated with a greater RH asymmetry [31].

A follow-up study found that MSR+ chimpanzees had increased cortical thickness bilaterally in the caudal anterior cingulate gyrus (mostly in the right hemisphere) and thinner cortex in the central portion of the pre- and postcentral gyri, primarily in the left hemisphere [32].

These results are too limited to draw distinct evolutionary patterns, but they are consistent with the RH asymmetry observed in humans in terms of SFR. That is, it is unclear whether the neural architecture that provides MSR in nonhumans is homologous to what occurs in humans and less clear whether there is a direct evolutionary path. However, taken together with what we find in humans, the data suggest that at least in chimpanzees, there are homologies rather than analogies. That is, the early data indicate that the same rightward bias for MSR that exists in chimpanzees also exists in humans.

#### **6. Functional Imaging Indicates Right-Hemisphere Dominance in Self-Face Recognition**

In the early 1990s through to today, research has examined how the brain actually allows MSR, treating it as an exceptional ability. Before neuroimaging, however, there were indicators of a possible right-hemisphere (RH) bias, as disorders of RH neural structures sometimes lead to a lack of own-body recognition [41,70–73] and disorders with selfawareness deficits appeared to be similar to RH disorder. Early attempts to determine the correlates of self-face recognition were made by pioneers such as Preilowski [74], who was the first to suggest a RH bias even though his methods involved indirect indicators.

In an early attempt to test Preilowski's hypotheses, we employed the WADA method in which the anterior portion of one hemisphere is anesthetized. Using self-face morphs (e.g., self-morphed to Marylin Monroe), it was found that following RH anesthesia, patients had significant difficulty recognizing their own face (Figure 1 [75]). There were early attempts to use lateralized hand response differences as a further test of RH SFR [65,66]. However, as is the case in much of cognitive neuroscience, fMRI (functional magnetic resonance imaging) dominates the literature.

**Figure 1.** Numerous methods have been employed to reach the conclusion that there is a RH lateralization in terms of MSR/SFR. Anesthesia applied to the brain in either the right or left hemisphere leads to differences in SFR. Namely, patients without a fully functioning RH see morphed images as not themselves. In this case, they report the image as Beyoncé under RH anesthesia conditions and their own face under LH anesthesia conditions [75].

Functional MRI (fMRI) of the self-face commenced in the mid-1990s with simple designs involving either unaltered self-faces contrasting with other faces or basic morphs [68]. Over time, the designs became more sophisticated, including examining affect, psychiatric disorders [76–80], and family [81]. While there is an overwhelming bias for RH activity in these studies, fMRI also revealed both a wide distribution of regions and the notion that many variables influence how the brain processes one's own face [82,83]. Much of the variability across the studies is for reasons unknown as the studies have not been replicated. Therefore, the fact that in one study a face is presented for a certain duration (for example) may be the factor, or it may be the task itself. We do find convincing data that suggest the more the SFR task engages self-reflection, the greater the bilateral medial frontal activity is [27].

Sugiura's group was one of the pioneers in determining the cortical correlates of SFR using fMRI [84–91]. Importantly, it was found that the brain has a number of distinct regions/networks associated with self–other distinctions. While there is overlap, the critical finding is that RH activation works in concert with medial frontal areas when SFR is performed in a social context. In other words, the social component of SFR appears to draw on the RH, as well as medial frontal networks.

Morita and colleagues also conducted numerous SFR fMRI studies and found a consistent RH bias [78,92–96]. In terms of brain symmetry, she solved a problem that has baffled researchers—the role that handedness plays in the lateralization of SFR [92]. They discovered that most right-handed participants exhibited a RH SFR bias. Likewise, most left-handers had RH bias, but there were significant numbers that had LH-localized SFR comparted to the right-handers. Therefore, it appears that, like handedness itself, which sometimes involves a shift in verbal language dominance, SFR may also shift to the LH in concert with some cases of left-handedness.

It may be surprising that three meta-analyses have been performed on SFR and the brain [27,97,98]. All three indicate a RH bias in SFR, though there are bilateral activations at a much lesser level. A key question that fMRI has helped to answer is the relationship between SFR and metacognition, either of oneself (SA) or of others (Theory of Mind: TOM). That is, if there is a relationship between SFR and TOM at a neurological level, such a relationship may be not just spurious but related in a meaningful way behaviorally and at an evolutionary level. It was found in two of the meta-analyses that SFR activates RH networks and overlaps with cortical midline structures in terms of metacognition, most likely critically involving medial regions of the frontal lobe [27,98]. We therefore conclude, based on fMRI, that the RH is likely necessary but not sufficient for SA.

Functional MRI provides only a correlational relationship between brain activity and behavior. By the 2000s, research was regularly appearing demonstrating a causal relationship. That is, by disrupting regions of the RH using noninvasive techniques, researchers were discovering that SFR was not just correlated with the RH but was actually involved in a causal relationship. The main methods employed thus far involve a version of a 'virtual lesion' in which a brain area is either temporarily taken offline or temporarily severely inhibited [99,100]. Basically, different regions of the brain were disrupted, and the subsequent changes in self-face recognition were measured, which established causality.

While early attempts using TMS (transcranial magnetic stimulation) did distinguish the right hemisphere as being necessary for self-face recognition [75], the most elegant of these studies was provided by Lucina Uddin and her colleagues [101]. She found that right parietal TMS disrupted self-face recognition, whereas left parietal TMS did not. Working in a somewhat 'backwards' manner, Uddin took these causal data and supported them (i.e., RH dominance) with correlational imaging data [102,103].

More recent studies have found that disruption of the RH is more dramatic for those individuals with subclinical grandiose narcissism [104]. That is, as narcissistic traits increase, TMS delivered to the RH causes a greater disruption of recognition of one's own face. This study, unlike studies of autism (see below), is one in which we see an excess of SA (rather than a deficit) correlating with some measure of SFR. We speculate that hyperactivity of self-associated neural networks associated with narcissistic traits leads to a steeper decline in function when disrupted (compared to normal activity), though this hypothesis needs further testing.

An even more interesting study revealed that the RH bias for self-faces may in fact be subconscious, below one even needing to identify whether the face is their own. Using a mental rotation task involving either one's own face or the face of another, Zeugin and his colleagues found that RH parietal TMS disrupted mental rotation of self-face compared to familiar faces in general [105]. This might indicate that the 'specialness' of the self-face is much 'deeper' in one's cognitive schema and does not rely on conscious representation. It would be interesting to test narcissistic traits (as in [104]) to see whether there are similar contributions of narcissism at the implicit level, as we previously observed at the explicit level.

Confirming these TMS studies in terms of brain/behavior disruption, a similar technique known as tDCS (transcranial direct cortical stimulation) was employed to alter the brain with self-face identification being measured. It was found that disruption of the temporo-parietal lobe in the RH disrupted self-face perception [106].

As is often the case in neuroscience, we have to ask whether we can extrapolate the data from the lab to the clinic to the 'real world'. While patient data from post-callosotomy (split-brain) surgery individuals [103] to those with autism [53,78,103] suggested a possible RH bias, a more specific method for detailing potential asymmetries is need. Specifically, we need to turn to those that lose self-face recognition in the absence of prosopagnosia (i.e., general face-recognition loss).

#### **7. Patient Data: Delusional Misidentification Syndrome**

While TMS and other neuroimaging points towards a RH bias in SFR, the data become infinitely stronger if matched within a patient population. Losing the ability to recognize the self is a hallmark of late-stage dementia. However, this deficit tells us little about the locality of SFR. Although rare, there are in fact cases where the loss of SFR occurs against the background of relatively stable cognitive processing [107].

Delusional misidentification syndrome (DMS) refers to circumstances in which patients form a fixed, distorted belief regarding the identity of a person, place, or object [107,108]. These disorders include Capgras syndrome, in which a person that was once very familiar to the patient (e.g., their husband) is now perceived as a stranger. Amazingly, there are cases where Capgras exists *exclusively* for the self, in which a patient misidentifies him/herself as being either a stranger they have not met or a different, familiar person that is not the patient. This disorder is rare, and there are only a few cases in the existing literature. It is not agreed upon yet what the naming of this disorder is, but it tends to be referred to as mirror sign or Capgras for the self [109–114].

In the most substantive review to this point, David Roane et al. [115] examined 24 case reports of the mirror sign. Of note is that within most of these cases, the patients were successful in correctly identifying the mirror images of *others*, signifying that they do not have a general impairment in recognition of familiar faces and they understand what a mirror does [116]. That is, the loss of face recognition was exclusive to the self and not due to prosopagnosia or a lack of understanding of what a mirror does/how it functions. In terms of localization, there was a diverse range of methods employed to obtain anatomical and functional data. Of the 24 patients, 9 of them had clear evidence for RH damage including the "parietal, temporoparietal, occipito-temporal, dorsolateral frontal cortex, basal ganglia, and thalamus" [115]. Out of 24 cases detailing mirror sign, imaging data were reported in 20 of them. Of the 18 MRI findings reported, 13 showed patients with mild generalized atrophy, as well as atrophy in specific regions within the right hemisphere of the brain. Further, PET and EEG findings supported RH dysfunction, displaying hypermetabolism in the right prefrontal, parietal, and occipital–temporal cortex and right temporal slowing. Thus, DMS can be considered a RH disorder [70–73].

In particular, a case report of the mirror image followed an elderly 77-year-old righthanded woman by the name of SP. SP was hearing impaired from a young age and was known to have communicated through sign language and also by lip-reading those around her. The patient's misidentification was regarded as highly selective as she was capable of readily identifying others in the mirror, though she regarded her own reflection as "the other SP", a companion of sorts. As expected, SP's lack of self-face recognition was supported upon her neurological examination; her MRI scan demonstrated clear RH damage [41].

The authors concluded that the association of RH involvement is consistent with previous work linking self-recognition to the right prefrontal and right frontoparietal cortex [42]. Overall, the most common findings were localized to the RH, which is not an unexpected finding [115].

#### **8. Why Does This Make Sense?**

The question remains as to why SFR is lateralized to the RH, and the clearest answer is that we do not know for certain. Most suggest that the RH has a specialization for social processing [117–122], though this ability is not exclusive to the RH. Rather, the RH (specifically the right TPJ) appears to be critical in correctly making self–other distinctions an ability needed for empathy and TOM [123,124]. In fact, it is now well established that the right TPJ is critical for 'feeling another's pain' and the ability to apply one's own feelings to another [123].

Overall, social patterns and self–other distinctions are biased to the RH, and it is plausible that SFR is tapping into this to the point that in the absence of SFR, there may be social deficits. For example, there are SFR deficits in pervasive developmental disorders [53,54,80,125–130], though the range of deficit is from severe to nondetected. Quevedo's group furthered this discussion by examining SFR in clinical populations including those with depression. Persons rated high in suicidal affect and cognitions, for example, have a unique neural response to the self-face compared to those without; self-faces involve differential neural circuits including the amygdala depending on clinical diagnosis and symptomatology [131–135]. These data indicate that SFR in its absence may indicate a lack of SA (i.e., autism) and that SFR in its presence may indicate different degrees of SA in different populations. In anorexia nervosa, a condition with a lack of accurate SA, there is difficulty in SFR [136]. Insight into one's own schizophrenia (i.e., schizophrenics that have awareness of their condition) correlates positively with SFR [137]. Further, disorders of consciousness also correlate (as indicated by physiological measures) with SFR [137].

The false belief tasks that are prevalent in TOM testing tap into the notion that the human brain is capable of modeling two brains/belief systems simultaneously [138–140]. This differentiation is critical for our social interactions as humans. For example, I need not be in pain to know that you are in pain. Parenting, reproducing, successful predation and predator avoidance, etc., are all enhanced by the ability to separate one's own thoughts from another's thoughts. Thus, and somewhat ironically, empathy and TOM are increased by the ability to separate one's SA from another's SA.

We propose that RH development and laterality has evolved for numerous reasons, though like others, TOM is likely the main impetus. What we propose, however, is that the role of teaching is overlooked in this domain. Humans are specialists at active teaching, and given the plasticity of the human brain, active teaching is a critical component of human survival. We examined the role of TOM in teaching and learning, as well as teaching communication, in a task that involved building simple Lego models and found that the best teachers have high TOM [141]. This task was demanding in terms of active teaching and social interaction, and of interest is that the learner's TOM was not critical (i.e., a learner can have low TOM and still learn), which models human–infant interactions.

Taken together, SFR appears to be an indicator of SA, and as SA fluctuates, so does SFR. RH patients are prone to deficits of SA [71–73,115,142]; thus, it is not surprising to see that SFR, SA, and the RH are related. It is worth noting that we do not know whether other nonape organisms achieve SFR, or even if SFR is possible with very different brain mechanisms or if one needs a highly lateralized brain to have SFR and/or SA [42]. Finally, we are still unclear as to whether SA is truly indicated by SFR. As it is, 30 years of neuroimaging and patient data have left us with more questions than answers.

**Author Contributions:** Conceptualization, A.J., B.B., M.P., K.C., J.B., N.A., S.Z. and J.P.K.; methodology; software, N.A.; validation N.A.; formal analysis, N.A.; investigation, A.J., B.B., M.P., V.M. (Victoria Mistretta), K.C., J.B., T.S. and J.P.K.; resources, N.A.; data curation, N.A.; writing—original draft preparation, A.J., B.B., M.P., V.M. (Vanessa Martinez), K.C., J.B., T.S. and J.P.K.; writing—review and editing, A.J., B.B., M.P., V.M. (Victoria Mistretta), K.C., J.B., T.S., V.M. (Vanessa Martinez), K.P., N.A., S.Z., A.S., S.S. and J.P.K.; visualization, A.J. and J.P.K.; supervision, J.P.K.; project administration, J.P.K.; funding acquisition, J.P.K., K.C. and J.B. All authors have read and agreed to the published version of the manuscript.

**Funding:** The work was funded by LSAMP (Louis Stokes Alliance for Minority Participation), The Crawford Foundation, and the Wehner Fund. Josh and Judy Weston provided funding as well as the Kennedy Foundation.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Not applicable.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


MDPI St. Alban-Anlage 66 4052 Basel Switzerland Tel. +41 61 683 77 34 Fax +41 61 302 89 18 www.mdpi.com

*Symmetry* Editorial Office E-mail: symmetry@mdpi.com www.mdpi.com/journal/symmetry

MDPI St. Alban-Anlage 66 4052 Basel Switzerland Tel: +41 61 683 77 34

www.mdpi.com ISBN 978-3-0365-7479-0