Recent trends in aroma release and perception during food oral processing: A review

Abstract The dynamic and complex peculiarities of the oral environment present several challenges for controlling the aroma release during food consumption. They also pose higher requirements for designing food with better sensory quality. This requires a comprehensive understanding of the basic rules of aroma transmission and aroma perception during food oral processing and its behind mechanism. This review summarized the latest developments in aroma release from food to retronasal cavity, aroma release and delivery influencing factors, aroma perception mechanisms. The individual variance is the most important factor affecting aroma release and perception. Therefore, the intelligent chewing simulator is the key to establish a standard analytical method. The key odorants perceived from the retronasal cavity should be given more attention during food oral processing. Identification of the olfactory receptor activated by specific odorants and its binding mechanisms are still the bottleneck. Electrophysiology and image technology are the new noninvasive technologies in elucidating the brain signals among multisensory, which can fill the gap between aroma perception and other senses. Moreover, it is necessary to develop a new approach to integrate the relationship among aroma binding parameters, aroma concentration, aroma attributes and cross-modal reactions to make the aroma prediction model more accurate.


Introduction
Human perception of aroma is a sense of smell that can arouse their memory and influence their mood and behavior.Consumers can satisfy their satiety and experience sensory and psychological pleasure during eating (Canon et al. 2018).Orthonasal perception (from the environment) and retronasal perception (from the oral cavity) perception are two ways of aroma perception.Importantly, the retronasal aroma perception plays the decisive role in human aroma cognition of different food due to the flavor characteristic perceptions of food consumption occurs in retronasal (Blankenship et al. 2019).Among the multiple sensory perceptions during food oral processing and aroma perception, the key factor in evaluating food sensory quality that drives food choices and consumer preferences, consists of multiple olfactory senses (Ployon et al. 2017;Pu et al. 2020;Spence 2021).Many researches have been conducted to elucidate the mechanisms of aroma release and perception through in vitro and in vivo methods.The research trends during last four decades  are summarized in Figure 1.Comparative analysis results (Figure 1A) of publication articles with the theme of aroma release and perception among different countries suggested that the overwhelming majority of articles mainly came from European countries, including France (170), England (79), America (65), Spain (56), Netherlands (51), United Kingdom (50), Germany (49), Italy (45), Australia (31), Japan (29), China (25), New Zealand (24) and Switzerland (17).Visualization analysis results (Figure 1B) showed that researches focused on the aroma release patterns in the mouth or retronasal of gum or solution models before 2012.Among them, the sucrose solution and limonene and ester compounds were most popular models.Subsequently, flavor evaluation of specific food consumption including wine, milk, and the other food products is predominant from 2014 to 2018.More recently (from 2018 to now), the dynamical sensory evaluation (especially the bread and wine perception), the identification of the olfactory receptors and their odorant activator and the behavior of consumer during food oral processing are becoming the research hotspots.The trend changes of these researches suggesting that the human perception and behavior is increasingly concerned.Overall, the aroma release and perception run through the whole research stage of food oral processing suggesting the importance of aroma perception.The retronasal aroma perception differs from the classical analyses of volatiles.The former provides information on either the total aroma composition in food or in air and aids in evaluating the sensory quality of food prior to eating.
The dynamic aroma release in the retronasal cavity during food oral processing is the in situ aroma perception for the profile of odorants transported from the oral cavity to the retronasal cavity, and it further binds with the olfactory receptor to generate aroma perception via the retronasal route (van Eck and Stieger 2020).Therefore, the aroma released during food oral processing is affected by many factors including food texture, food composition properties, and oral parameters.It was reported that the release of tastes compounds (salts, sugars, acids, etc.) (Pu et al. 2021a), the change of food texture (microstructure, specific surface, viscosity, etc.) (Bonneau et al. 2018;Pu et al. 2021b), and the parameters of oral processing (saliva, chewing frequency, and oral mucosa) (Canon et al. 2018;Ployon et al. 2020) all could affect the rate of aroma release during food oral processing.Furthermore, there were cross-modal relationships among aroma, taste, and texture perception.Individual differences largely influence the aroma perception, such as differences in personal experience and cognition, genetic, and health status.Among them genetics and health status are the decisive influencing factors.For example, people with genotype variation of TAAR5 cannot detect or discerned the fishy smell but recognized it as the potato, caramel, or rose (Gisladottir et al. 2020).Besides, the obese people or patients people with olfactory system disorders have higher threshold values of odorants (Jacobson et al. 2019), or even the COVID-19 patients have already lose the sense of smell (Brann et al. 2020).
Although there are challenges in the research of aroma release and perception during food oral processing, many analytical methods have been proposed to overcome them.First, with the development of in vivo and in vitro extraction methods and the on-line analytical instruments, a deeper understanding of the aroma release from food to human perception has been developed.Second, different characteristics of foods, model solutions/gels, and chewing simulators are proposed to investigate the effective factors and their impact mechanisms on food property and oral parameters on aroma release (Genovese et al. 2015;Peyron et al. 2019).Third, biotechnological processes, such as in vivo imaging, calcium imaging, heterologous expression, and high throughput screening of dual luciferase, can help in characterizing the specific olfactory receptors (ORs) and confirming their corresponding active aroma compounds (Trimmer et al. 2014;Ahmed et al. 2018).Fourth, several interdisciplinary subjects are further applied to solve the mechanisms of aroma release and perception, such as chemistry, biology, materials, engineering, mathematics, physics, and even medical science (Chen 2009).
Clarifying the mechanisms of aroma release and perception during food oral processing is of great contribution to control aroma release, enhance the characteristic aroma, reduce the off-flavor, and improve the sensory quality of healthy food.This work discussed and summarized the research progress during the last four decades, and current challenges of aroma release and perception during food oral processing.This review hopefully could provide a reference for producing better flavored food, meeting the preferences of consumers, and prospecting the development direction of aroma release and perception.

Physical and chemical properties of aroma compounds
Aroma compounds with low molecular weight (<400 Da) and boiling point (<200 °C) get easily released in gaseous phase from the food matrix and further get perceived by the ORs (Dunkel et al. 2014;Xu et al. 2022aXu et al. , 2022b)).The Log P value (log of octanol/water partition coefficient), partition coefficient and Henry's Law are important parameter to assess the relative concentration of aroma compounds with being released from liquid, oil, or complex food matrix in the oral cavity (Lesme et al. 2020a;Pu et al. 2020).The partition coefficient of each odorant is constant at a fixed temperature and in aqueous solution (Lafarge et al. 2008;Pagès-Hélary et al. 2014), but it can be dynamically changed due to the changes of the complexity food-saliva matrixes during food oral processing (Heilig et al. 2016).The contribution of aroma compounds to the overall aroma profile depends on the odor activity value (OAV, the ratio of concentration to odor threshold).Currently, 226 aroma-active compounds with OAV over 1 were identified as the key odorants among 227 different foods (Dunkel et al. 2014).Importantly, the chemical structure properties, including functional groups, carbon chain, unsaturated, carbon-carbon double bonds, additional methyl group in the carbon chain, and chiral conformation, play the key role in determining the aroma attribute and odor threshold (OT) value.For example, most of the sulfur and nitrogen containing odorants had the lower OT but their olfactory power would be reduced after blocking the SH group by alkylation (Tan and Siebert 2004;Polster and Schieberle 2015).The OT of alcohols decreased along with the chain length but it was contrast in the ester compounds (Sowndhararajan et al. 2015).The OT of (S)-sotolon (1.7 µg/kg) was significantly lower than that of the (R)-sotolon (87.0 µg/L) due to different chiral chemical structures (Haag et al. 2021).Currently, the 3D-QSAR, artificial neural network, and machine learning are most popular prediction methods for screening the target aroma compounds according to the physical and chemical structure properties (Polster and Schieberle 2015;Kowalewski and Ray 2020).

Food composition
The protein, fat/oil, carbohydrate, mineral and other components in food could modify or affect the aroma release from the food matrix to the nasal cavity during oral processing.Generally, higher fat content slows down the aroma release rate, resulting in longer time for aroma perception to reach its maximum intensity (Feyzi et al. 2020;Zhou et al. 2021).Especially, the aroma release such as limonene, 2-heptanone, ethyl butyrate, and ethyl hexanoate can be significantly inhibited in the emulsion system with higher fat content (Tarrega et al. 2011).The protein adherence to oral mucosa and epithelial tissue could change aroma perception by binding with the aroma compounds (Lesme et al. 2020b).Thus, increasing the proportion of protein in food tends to reduce the release of volatile which impacts the overall aroma perception (Zhang et al. 2021).In most cases, carbohydrate like sucrose can promote the aroma release during chewing gum (Yang et al. 2017).According to the cross-modal interaction between taste and aroma perception, when the flavor quality is consistent in aroma and taste, taste substances can compensate for aroma perception.For example, the minty, vanilla, and cherry aroma could increase the sweetness intensity when chewing gum or consumption of sugar solution (Konitzer et al. 2013).Sodium chloride contributes to saltiness perception and plays a significant role in improving flavor and promoting aroma release during food oral processing (Pu et al. 2021a).

Food texture
Food texture is well known as a sensory property that affects enjoyment and preference of food due to the cross-modality relationship between texture and aroma perception during oral processing (Bonneau et al. 2018;Lesme et al. 2020b;Guo 2021).Many food oral processing models have been developed such as mastication of the yogurts, bread, cheese, potato chips, gum, pectin, and gel.The gel model demonstrated that soft gel could release higher amounts of aroma with faster release rate of aroma than that of hard gelatin (Yang et al. 2017;Ćorković et al. 2021).However, it is the opposite in fruit that the mango pulp with higher viscosity was not beneficial for the rapid release of volatiles (Bonneau et al. 2018).The food particle size also affects the release rate of aroma compounds during solid food oral processing.In our previous work, we have confirmed that the bread microstructure determined the oral processing behavior, affecting the aroma perception (Pu et al. 2019b(Pu et al. , 2020(Pu et al. , 2021)).Raithore and Peterson (2018) demonstrated that during the gum chewing with the average particle size of 57 μm showed the higher aroma release rate than that of 246 μm.The crispness level also affected the dynamic aroma perception during the consumption of potato chips (Luckett et al. 2017).In addition, the liquid foods are also an important part because the interfacial mass transfer affects the aroma release directly, and the different release kinetics can be influenced by the property of aqueous solution (Mao et al. 2017;Weterings et al. 2020).The Stokes-Einstein equation confirmed that diffusion is negatively proportional to viscosity (Edward 1970), that is, the increase of viscosity resulting in decrease of aroma release (Bayarri et al. 2006).However, the emulsion ingredients and structures undergo significant changes during emulsions consumption, resulting in different aroma release kinetics and perception intensity.Benjamin et al. (2012) reported that the micrometer or submicrometer range droplet size had the fast aroma transfer kinetics, which was different from the work of van Ruth et al. (2002) which reported that smaller droplets with larger interfacial area slowed aroma release and reduced air-liquid partition coefficients.These findings elucidate that the multiple properties of emulsion have a multi angle effect on the aroma release which should be further studied.

Internal influencing factors
As the first step of food digestion, food oral processing is important for nutrient intake and the sensory enjoyment of food.When foods are consumed, they are broken down into small particles by the synergy of various organs, including the lips, tongue, and teeth.The bolus formation is performed by saliva, postcanine teeth, tongue, and checks.The saliva can increase the lubrication and cohesive property of the food bolus for its safe swallowing (Chen 2009).The varying aroma release patterns in the oral and nasal cavity could be due to the difference in ethnicity, gender, and physiological parameters (Pedrotti et al. 2019).Therefore, the oral physiological parameters such as bite force and chewing frequency, saliva content and composition, oral mucosa, or air flow significantly influence the aroma release and perception during food oral processing.

Mastication
Mastication is based on the eating experience and is easily influenced by the texture properties and composition of food.Chewing efficiency depends on the bite force and the manner food broken down, and significantly relates to the physiological characteristics such as sex, age, ethnic group, and living habits of an individual.Pressure film force sensors and surface electromyogram are the main instruments to measure bite force and analyze electrical signals (that emanate during muscular contraction), respectively (Yen et al. 2015;Pu et al. 2021b).Generally, males have a greater bite force than females.Besides, ethnics with higher meat eating tendency such as the young adults of Mongolian had significantly higher incisor bite force (168 N) than those of Han (146 N) in China (Ma, Zhang, Hu, Wilde, and Chen 2021).Release rate of odorants from food matrixes differs due to different chewing rates (frequency) of food in the mouth (Luckett and Seo 2017).Genovese et al. (2014) also confirmed that consumption of semi/solid foods with multiple small bite sizes or longer time of mastication and liquid foods with multiple sips could promote the efficiency of food oral processing.Therefore, a high frequency of chewing results in smaller particle size and a larger contact surface promotes the release of more aroma compounds into the nasal cavity (How et al. 2021).Additionally, the longer the oral chewing time, the longer the transmission time of the aroma compounds in the oral cavity, resulting in a higher intensity of aroma perception (Okawa et al. 2021).

Saliva
Saliva derived from the submandibular (65%), parotid (20%), sublingual (8%), and numerous minor glands (< 7%) is a complex biological fluid with multiple functions (lubrication and protection, buffering and clearance, antibacterial activity, maintains the tooth integrity, and flavor perception) (Liu and Duan 2012;Javaid et al. 2016;Pedersen et al. 2018).Saliva with pH of 6.0 ~ 8.0 is mainly composed of water (99%), salts, and proteins along with the food residue, cellular debris, and microorganisms (Canon et al. 2018).The saliva plays a key role in protecting, lubricating, and promoting the formation of bolus, and the transportation of flavor compounds (Chen 2009;Ployon et al. 2017;Canon et al. 2018).Due to the varying food stimulation, the changed saliva secretion, the salivary protein composition and the complex microbial environment regulate the dilution and concentration of aroma compounds at their release (Schwartz et al. 2021).Doyennette et al. (2011) investigated the release of diacetyl at different concentration of glucose solutions during oral processing by experimental and modeling approaches and the in vivo tests confirmed that the aroma dilution factor by saliva were approximately 10%.The saliva flow can help dilute the aroma retention effect of mucous membranes after swallowing (Buettner 2002), and it primarily influenced the compounds with a high hydrophobicity (Khramova and Popov 2022).Of note, the aroma release is related to the variability of complex saliva composition (Tarrega et al. 2019).
More than 3,000 salivary proteins and peptides are identified by proteomic platforms, where most of them are of microbiological origin (Castagnola, Scarano, Passali, Messana, and Paludetti 2017;Schwartz et al. 2021).Most of these proteins are enzymes with catalytic ability (oxidation, reduction, hydrolysis, etc.) or binding ability (non-covalent or covalent binding) affecting the aroma release and perception during food oral processing (Ployon et al. 2017).Esters could be partially hydrolyzed (esterases) in human saliva to form the corresponding alcohols (María et al. 2020).Thiols and aldehydes could be easily oxidized (peroxidases) and reduced (dehydrogenases and aldehyde reductases), respectively (Buettner and Andrea 2002).Moreover, the antioxidant capacity of saliva also had a significant effect on the retronasal aroma perception (Muñoz-González et al. 2020).During food oral processing, the saliva-food system was changed due to the binding of salivary proteins like salivary proline rich proteins, mucins, and α-amylase with the polyphenols or tannins so as to affect the aroma release (Genovese et al. 2015).Therefore, the aroma release significantly depends on the properties of saliva.For example, the obese people with lower activity of α-amylase reduced their sensitivity to food aroma by inhibiting the release of volatile compounds during wine consumption (Piombino et al. 2014).As of now, 11 enzymes/proteins (mucin, α-amylase, peroxidases, lactoperoxidase, esterases, lipase, lipid oxidation enzymes, carbonic anhydrase Ⅵ, lysozyme, dehydrogenases/ aldehyde dehydrogenases, and superoxide dismutase) have been reported to catalyze the changes in aroma via complex physicochemical reactions such as hydration, hydrolysis, oxidation, or interaction during food oral processing (Pagès-Hélary et al. 2014;Ployon et al. 2017).Thus, the model experiment of saliva-food interaction should use fresh and whole saliva rather than the pretreated human saliva (frozen, heated, or centrifuged) (Muñoz-González et al. 2018;Ployon et al. 2017).

Oral mucosa
Aroma compounds are perceived during food consumption and dissipate rapidly after swallowing.The speed of aroma dissipation in the mouth determines the sensory quality of food.On the basis of its structure, the oral mucosa differently affects the aroma perception during chewing and after swallowing.Although mucin located at the surface of the oral mucosa has an important effect on the aroma persistence due to the well-binding capacity of mucin interactions with the odorants, the mechanisms of aroma retention by oral mucosa are still unclear.In vivo experiments combined aroma residue detection by solvent extraction of exposure saliva (Buettner 2002), modified stir-bar sorptive extraction (Buettner and Welle 2004), and solid-phase microextraction (SPME) (Esteban-Fernãndez et al. 2016) elucidated that the aroma compounds could bind on the surface of the oral mucosa and gradually release into the oral cavity modifying the aroma release and perception.
Moreover, the proton-transfer-reaction mass spectrometry (PTR-MS) combined with the oral mucosa cell-based model tests confirmed the surface properties of mucosa cell was changed and exposed the binding sites of aroma compounds, impacting the partition of odorants, and even metabolize/ degrade them during oral processing (Ployon et al. 2020;Muñoz-González et al. 2022).To address the issue of aroma loss through ingestion the biopolymer pullulan that could bind to oral mucosa and controlled aroma release by salivary α-amylase was developed (Dinu et al. 2019).

Nasal airflow
Nasal airflow changes the aroma concentration received by olfactory epithelial cells, and affects the processing of olfactory perception by the central nervous system of olfactory cells (Yao et al. 2020).Therefore, understanding the nasal air flow can help us better design food formulations during oral processing.Three representative analytical methods are summarized in investigating the effect of nasal airflow on aroma perception: (1) Constructing the theoretical model.The theoretical simulation model of nasal air flow during the consumption of aqueous solution and gum was calculated based on the oral physiological structure and aroma partition coefficient in saliva (Figure 2A) (Trelea et al. 2008).
To accurate the theoretical model the raw data are corrected by using the signal from acetone in the respiratory airflow (Normand et al. 2004); (2) Directly determination of the air flow.The aroma perception within different nasal air flow and air pressure could be directly determined by rhinometer (Yao et al. 2020; Figure 2B).Moreover, the mechanical vibration caused by airflow participated in the regulation of the central nervous system and promoted the aroma perception by the unilateral comparison of olfactory test airflow on both sides of the nasal cavity (Yao et al. 2020).Currently, laser doppler anemometry (LDA) is one of the most accurate analyzer in nasal air flow measurement.The nasal airflow could be simulated (Figure 2C) in less than five minutes based on the LDA method (Berger et al. 2021); (3) Imaging and fluid mechanics combined with 3D printing.The transportation of aroma compounds to the retronasal cavity was a series of alternating dynamic and static events due to complex oral processing movement including opening/closing of mouth, mastication, and breathing.Magnetic resonance imaging (MRI) and computed tomography (CT) imaging techniques are important methods to determine the correlation between specific anatomical areas and olfactory evaluation performance by comparing the individual nasal anatomy (Figure 2D).Computational fluid dynamics (CFD) technology can quickly convert CT scan of the nasal cavity into an anatomically accurate three-dimensional numerical model.Thus it can be used to predict airflow and odorant transport, and finally determine olfactory sensitivity.The oropharynx anatomical structure formed an air curtain outside during respiration, preventing aromatic compounds from being transported to the mainstream toward the lungs, was confirmed by 3D printing experimental model (Figure 2E), CT image and airway flow field analysis (Ni et al. 2015).This result provides the first convincing evidence of the adaptability of human oropharyngeal geometry to effectively transport odorants to the olfactory receptors in the nasal cavity.To better understand the mechanisms of nasal air flow on the aroma perception, CT and MRI combined with the specific software to obtain the valuable 3D reconstruction (Figure 2F) of the nasal cavities help us to accurately understand the structure of the nose and the transportation of nasal airflow during breathing.

Analytical methods
Food aroma perception is resulted from the interaction between food and human.Due to the complex oral processing procedures, the oral and nasal cavity aroma profiles significantly differ from food matrixes.Therefore, the clarification of the mechanisms of aroma release and the perception during food oral processing is required.Many enrichment methods, analytical instruments, and dynamic aroma perception evaluation methods have been proposed to elucidate the aroma change patterns in oral and nasal cavity.
With the development of the gas chromatography (GC) technology since the 1960s, GC-flame ionization detector (GC-FID), GC-mass spectrometry (GC-MS) combined with different extraction methods such as SPME, stir bar sorptive extraction (SBSE), Tenax in low-temperature extraction, syringes cold trap, headspace extraction (HS), and Purge & Trap (P&T) have been widely developed, respectively.Therefore, the lower concentration of aroma components released in the nasal cavity could be detected (Figure 3, the supplemental data presented in Table S1).Figure 3(A) shows that as for the traditional analysis methods, the most volatile compounds are detected by SPME (66), followed by Tenax (63) and HS (44), far more than other methods.Although many aroma compounds have been identified from the oral or nasal cavity using these analytical methods, some shortcomings during the analysis require attention.First, the HS-GC-MS method has a high detection limit.The aroma compounds released during oral processing are significantly lower than that in the food matrix.Second, SPME, SBSE, P&T, and Tenax methods require a long period for equilibrium and extraction.However, food processed in the oral cavity dynamically release their aroma within a few seconds, such as drinking of wine, soup, and coffee, or consumption of solid/semisolid food.In the long-term equilibrium or extraction process, saliva will continue to immerse food masses and affect their texture properties, meanwhile the salivary proteins and enzymes mentioned above will affect the release of aroma components (Salles et al. 2007).Moreover, these extraction methods do not provide an easy and direct collection of the volatiles during food oral processing.Most of these methods detect the food pellet or flavor model after chewing, that has a certain gap from the actual dynamic change process.
In view of these shortcomings, three on-line analytical instruments namely atmospheric pressure chemical ionization-MS (APCI-MS), selected ion flow tube-MS (SIFT-MS), and proton-transfer-reaction mass spectrometry (PTR-MS) with high sensitivity (pptv), convenience (without pretreated), and rapid (real-time monitoring) properties have been developed.The PTR-MS/PTR-TOF-MS analytical instrument was invented for monitoring aroma in air with lower concentration (Lagg et al. 1994;Hansel et al. 1995;Jordan et al. 2009).This instrument is widely used to monitor the real-time change in aroma compounds in environmental sciences and food industries such as the processing of banana, apple, wine, bread, and dairy products (Hageman et al. 2019).
The SIFT-MS analytical instrument was invented for medical research (Adams and Smith 1976;Smith et al. 1996), such as analyzing breath volatiles in patients with different diseases and the dynamic monitoring for specific medical biomarkers (Smith et al. 2008).Subsequently, it was applied in food flavor chemistry, such as monitoring the aroma release during the process of chewing garlic, mint, tomato, and strawberry (Langford et al. 2019).The APCI-MS analytical instrument was invented (Linforth and Taylor 1999) for analyzing aroma release during food oral processing or in other chewing models, like strawberries, peppermint, gum, and tomatoes (Raithore and Peterson 2016;Genovese et al. 2019).More recently, GC-ion mobility spectrometry (GC-IMS) with GC separation and high-resolution ion migration technology was commercialized, to achieve efficient secondary separation of the two-dimensional approach.In our previous work, the aroma release in mouth and its transport to the nasal cavity during white bread consumption has been monitored by BreathSpec® GC-IMS (Pu et al. 2019a(Pu et al. , 2020)).It was found that the key odorants perceived in the retronasal cavity were different from the bread matrix and their concentration was about 10-fold lower than that from the oral cavity.
The detection results showed (Figure 3A) that PTR-MS/ PTR-TOF-MS identified 108 aroma compounds, followed by APCI-MS (50), SIFT-MS (38), and GC-IMS (19).A total of 221 aroma compounds were reported during food oral processing (Figure 3B).Among them, esters were predominant (52), followed by aldehydes (36), ketones (35), alcohols (31), and nitrogen compounds (20), and the other compounds were less than 20.Although many aroma compounds have been identified, the key odorants contributing to the retronasal aroma perception lack deep investigation because the aroma profiles dynamically change and short process time in aroma release during food oral processing.Besides, the key odorants percieved in retronasal cavity differ from that of real foods (Pu et al. 2019a(Pu et al. , 2020)).Therefore, identifying these key odorants and establishing their fingerprint from retronasal perception could help in better understanding the aroma release and perception.

Artificial mouth simulators
In vivo experiments have figured out that the inter-individual variations and the ethical compliances were great challenges for decoding the aroma release during food oral processing (Panda et al. 2020).Therefore, developing an in vitro approach is an effective alternative.Previous studies have demonstrated that individual differences impacted aroma release and perception patterns due to the variances in oral parameters such as mouth volume, saliva, bite force, and chewing efficiency (Buettner 2002;Muñoz-González et al. 2019).To eliminate the interference of food properties and mastication parameters during oral processing, several in vitro simulation conditions (artificial saliva) and different artificial mouths (chewing simulators) have been developed and employed so as to stimulate and predict the aroma compounds during their consumption for developing better foods (Salles et al. 2007;Peyron et al. 2019;Panda et al. 2020).For the liquid samples, including wine, olive oil, whiskey, and flavoring model, the simulated mouth conditions and retronasal cavity combined with the aroma monitor instrument are the most popular methods to investigate the effect of saliva on aroma release (Itobe et al. 2015;Genovese et al. 2015;Esteban-Fernãndez et al. 2016).Drinking volume like smaller sips of coffee reduced aldehyde, while β-damascenone and 4-vinylguaiacol increased during coffee consumption (Genovese et al. 2014).The impact of saliva also depends on the volatile (concentration and hydrophobicity) and nonvolatile composition of foods (Piombino, Moio, and Genovese et al. 2019), thus elucidating that the aroma released from the food matrix, bolus, or saliva-liquid matrix changes dynamically.Therefore, it is necessary to develop more accurate and suitable equipments for simulating the procedures and situation during oral processing.Many artificial mouth mimic different status foods oral processing including liquid, semi-solid, and solid have been proposed, in this review four representative simulators are listed as follows: (1) Equivalent simulated chewing equipment.Krause et al. (2011) invented an equivalent chewing device to simulate the breaking down of apple and gum during oral processing (Figure 4A).However, the human masticatory system is comprised of the upper and lower jaw with teeth and the bite force, and the chewing frequency, and time dynamically change during chewing according to the texture property of the bolus (Chen 2009;Guo 2021;Pu et al. 2021).( 2) Though the solid food could be similarly broken down into small particles like human mastication, the formation of the bolus was significantly different from chewing in mouth.Besides, during food consumption, the aroma is gradually released from food to the oral cavity and then transferred to the retronasal cavity by the nasal air flow.Therefore, a chewing simulator (Figure 4B) including the chewing system (upper and lower jaw with 3D printing teeth, tongue stir, oral cavity), the gas flow system (simulates the nasal flow), the mouth temperature, and the saliva flow were developed (Salles et al. 2007).The several mobile parts of this chewing simulator can accurately reproduce shear, grinding strength of teeth, stirring function of tongue, as well as the aroma release in real-time.(3) To overcome the difficulty in the formation of food bolus and to present a more realistic simulation of food aroma release, Woda and Peyron (Peyron et al. 2019) developed an artificial masticatory advanced machine (AM2) (Figure 4C) to simulate the texture property of beef, sausage, coconut, pasta, gherkins, pork meat, peanut, and carrot during oral processing.It simulates the chewing behavior among people of different ages and produces food bolus with similar granulometric characteristics.( 4) With a rapid growth in simulation and 3D printing technologies (Alemzadeh et al. 2021), the same-size ratio of artificial mouth with intelligent control and learning functions can simulate the aroma release and transfer mechanisms during food oral processing more realistically (Figure 4D).Finite element models can efficiently account for oral processing parameters, through the simulation of food deformation and fracture during the first bite.Skamniotis et al. (2019) also proposed that finite element models could efficiently account for oral processing parameters by the simulation of food deformation and fracture during the first bite.

Mechanisms of olfactory signal transduction
Smell is perceived by orthonasal olfaction via the nostrils and retronasal olfaction assisted with nasal airflow.Due to the differences in physiological structure and oral mucosa of human cavity and nasal the threshold values of odorants in orthonasal and retronasal differed (Heilmann et al. 2004(Heilmann et al. , 2008;;Small et al. 2005).Moreover, the retronasal olfaction in non-equilibrium situation is affected by the complicated food-saliva environment, which is evoked by multiple neural responses such as taste, touch, and trigeminal nerve (Small et al. 2005;Piombino et al. 2014).In Figure 5, the main olfactory system consists of subsystems with distinct anatomy, olfactory receptor repertoires (olfactory sensory neuron, OSN), signal transduction mechanisms (cribriform plate and olfactory bulb), and central projections (cerebral cortex) (Trimmer et al. 2017).The main function of subsystems is to detect the variances in aroma compounds from foods or environment and subsequently transduce them into neural signals.First, the aroma compounds reversibly bind to the odorant-binding proteins (OBPs) across the aqueous nasal mucus (Gonçalves et al. 2021).The main function of OBPs is to facilitate the delivery of hydrophobic aroma compounds across the aqueous mucus (Briand et al. 2002;Gonçalves et al. 2021).Human OBPIIa is a monomeric protein that exhibited the classical lipocalin fold with a conserved eight-stranded β-barrel, harboring a remarkably large hydrophobic pocket (Schiefner et al. 2015), which has high binding energy on binding the pleasant odorants than unpleasant (Castro et al. 2021).Second, the loaded OBPs transport the odorants to bind with the olfactory receptors (ORs) for detecting odorants dissolved in mucus and bonded with OBPs.
The ORs, expressed in the cilia of OSN, are a large multigene family (855, accounting for 2-3% of all genes) of G-protein-coupled receptors having seven-transmembrane domain topology.They play a critical role in cell recognition, activation of signal transduction, and regulation of sensory information (smell, taste, pain, and vision) (Keller et al. 2007).The OR gene is highly variable but only 350 ORs code for functional receptors (Gisladottir et al. 2020).The mammalian ORs are categorized according to their sequence homology: class I and class II.Class I is more sensitive to water-soluble odorants and is the only heterologous receptor present in fishes.Class II is sensitive to hydrophobic odorants that 90% of mammals account for.
In addition, there are six intact members in human TAARs that respond preferentially to amines (Gisladottir et al. 2020).When the odorants bind to the ORs or TAARs, a typical cyclic adenosine triphosphate (cAMP) signal transduction cascade is initiated, finally the neurotransmitters are released at the synapses of the main olfactory bulb (Small et al. 2005).The OR activates its associated G-protein-Gαolf and causes conformational changes in GPCR, including tilting and rotation of TM6 relative to TM3.The G-proteins, a member of the Gαs subfamily, contain three subunits (α, β, and γ).Among them Gα-olf is considered as the active unit that is specific to the ORs while β and γ subunits regulate the activity of the Gα-olf.The activated Gαolf can subsequently activate the adenylyl cyclase III (ACIII) after the dissociation from its β-and γ-subunits.The production of cAMP is stimulated by ACIII, that binds to odorants and opens a cyclic-nucleotide-gated channel (CNGA2).CNGA2 initiates the depolarization of the neuron and terminates the release of neurotransmitters at the synaptic terminal in the main olfactory bulb (MOB) (Trimmer et al. 2017).Each ORN expresses the same OR which is projected onto a specific OB and is equivalent to an integrated processing of the signals input.
The aroma perception can be finally formed when the electric signals are encoded with MOB transmitting to the central olfactory system (Figure 5D).However, the olfactory neural pathway and the aroma perception mechanisms are still unclear.The encoded aroma signals can simultaneously activate the multiple areas of brain center, including the piriform cortex (PC), amygdala complex and entorhinal cortex (Anderson et al. 2003;Trimmer et al. 2017).The active pattern of OB activation of the PC is randomly (Yamada et al. 2017).The amygdala encodes the intensity of aroma and the quality of aroma characteristics and also involves emotional regulation.The entorhinal cortex can feedback and regulate olfactory perception.Among the multiple areas of brain center, the PC is the key node of the cortical olfactory system.Porada et al. (2019) have confirmed the PC play the key role in regulating and coding the cross-modal interaction including the smell, taste, vision, and sound.Then the olfactory information is conveyed to the brain areas including the orbitofrontal cortex (OFC), insula, thalamus, hippocampus, ventral striatum, hypothalamus, cingulate cortex, and the cerebellum for the perception of odorants (Freiherr 2016).The classic model believes that multi-sensory synergy occurs in the OFC area, which not only deals specifically with odor perception, but also participates in important processes such as perceptual decision-making and distinguishing perceptual uncertainty (Iravani et al. 2019).The intensity of response signal in OFC region of cerebral cortex can be used as an effective method to evaluate the olfactory cognition and perception in clinical diagnosis.The cerebellum plays an important role in regulating olfactory processing, and adjusting the smell amplitude according to the aroma concentration.The insula is traditionally considered as a primary gustatory cortex and receives input from not only PC and amygdala but also OFC.Additionally, the cingulate cortex the limbic system also participates in the information integration of the olfactory-taste cross-modal role (Patin and Pause 2015).These research progresses illustrate the basic process of olfactory cortical system forming aroma perception and the relationship with other sensory perception.

Olfactory receptor and binding mechanisms
Three characterization methods are proposed because the data on specific olfactory receptors and their ligands are incomplete.First, heterologous expression of the ORs in Hana3A or HEK-293 cells, followed by a high-throughput assessment of the OR activation on stimulation by odorants using the luciferase reporter assay to confirm the specific olfactory receptors (Trimmer et al. 2014).Using this method, a large number of ORs and effective combined ORs for the relationship among aroma synergy, inhibition, and enhancement have been found and confirmed, respectively (Singh et al. 2019).This method based on the cAMP-responsive reporter gene, allows for the measurement of OR activation through the canonical second messenger pathway.Furthermore, due to the overlap effect in response properties for olfactory neurons, it is advantageous over calcium imaging and electrophysiology.
Second, the method of anatomy.The olfactory nerve transduction mechanisms allow receptor specific axonal inputs to coalesce into glomeruli of the olfactory bulb.McClintock et al. (2014) screened the eugenol and muscone ORs in vivo based on the activity dependent expression and the expression of one OR gene per olfactory sensory neurons.Generally, the interactions of odorants at receptors modulate the encoding of odor signals.March et al. (2020) identified 30 responsive ORs by determining OR response patterns of 4 odorants and 3 binary mixtures in vivo in mice.In addition, the technical obstacles presented by the anatomy of OB, prevent successful imaging of its medial region.
Third, in vivo imaging technique (Shirasu et al. 2014).The genetically encoded probes have great advantages.For example, OMP-spH mice expressed spH under the control of an olfactory marker protein promoter in all mature olfactory receptor neurons (Bozza et al. 2004).On the basis of retrograde labeling of neurons, Ca 2+ imaging and single cell reverse transcription PCR, it is possible to reverse search for the olfactory neurons by imaging the olfactory bulb that could be anterodorsomedially present in the OB.Using retrograde tracing techniques, one can confirm a specific neuron cell and its corresponding olfactory receptor.Using in vivo imaging technique, the mouse musk receptor MOR-215-1 and human musk receptor OR5AN1 were confirmed (Ahmed et al. 2018).Generally, the heterologous expression would further validate the OR in humans, even though the second and third methods have confirmed the target ORs.
Till now, a total of 112 human ORs and their corresponding activated aroma compounds (ligands) have been reported (Figure 6).Only 90 ORs and 196 corresponding odorants with 50% of maximal effect concentration (EC50) are identified (Figure 6A).Among these odorants ketones (59) were the most identified ligands, followed by esters (31), aldehydes (29), alcoholics (19), and phenolics (15).However, identification of the odorants that can activate ORs is lacking due to the diversity of ORs (around 400) and each of them could be activated by multiple odorants.Therefore, it is necessary to identify the specific ORs and their ligands (with EC50 value) for further decoding the aroma binding modes so as to elucidate the aroma perception mechanisms of cooperation and inhibition.As the crystal structure and functional data of human ORs are still unknown, analysis by computer modeling is suitable for delivering reliable propositions on the structure of the ligand-binding sites and specific ligand-binding modes of particular ORs providing a deep insight into odorant selectivity.Homology model of ORs; quantum mechanics and molecular mechanic calculation; molecular dynamics simulation; crucial site; and 3D-quantitative structure-activity relationships are five main steps for systematic analysis of the binding mechanisms.

Sensory evaluation of aroma perception
Sensory evaluation is an effective tool in improving the sensory and quality control of foods.Aroma perception results from physical and chemical stimulation, physiological response, and psychological effect of aroma compounds.The representation of olfaction is mainly reflected on aroma characteristics and intensity.Moreover, aroma evaluation is also a bridge between the volatiles and aroma characteristics.The aroma perception during food oral processing is a dynamic process.Therefore, the key to evaluate aroma quality of product and consumer preferences is to identify and quantify the changes in perception aroma attributes and to directly distinguish the aroma perception of different food during oral processing.
According to the basic principles and characteristics, four classes of sensory evaluation methods are summarized, including intensity comparison method, contour comparison method, classification, and multi-dimensional comparison method.(1) Intensity comparison method.The time-intensity method (TI) (Luckett and Seo 2017), intensity scale (IS) (Cruz et al. 2013), and dual-attribute time-intensity (DTI) (Findlay 2017) require panelist to score the overall strength of one or two aroma characteristics on the time axis.However, these methods are difficult to change the overall intensity and the large individual difference using a standard reference as a measurement scale.(2) Contour comparison method.The quantitative descriptive analysis (QDA) (Albert et al. 2011), rapid sensory profiling (flash profiling, FP (Liu et al. 2018), projective mapping (Napping®) (Antúnez et al. 2017), and time scanning descriptive analysis (TSDA) (Antúnez et al. 2017).The latter three methods are evolved from the QDA method, according to the actual requirements of different evaluation samples and the evolution of evaluation environment.The FP and Napping ® methods only require a fast evaluation ability familiar with the target flavor profile.The TSDA method is aimed at specific foods like tea, coffee, etc. as their flavor characteristics change significantly with temperature fluctuation.(3) Classification.It includes free multiple sorting (FMS) (Dehlholm 2015) and polarized sensory positioning (PSP) (Ares et al. 2018).FMS require the panel to classify the given samples according to personal experience.The PSP method classifies samples according to extreme differences (very similar or very different).They are widely used in enterprises, such as product imitation or product orientation modification.(4) Multi-dimensional comparison method combines flavor attributes and change characteristics to explain the differences in sample contour, and change rules from multiple perspectives.The temporal dominance of sensations (TDS) proposed by Pineau (2009) and temporal check-all-that-apply (TCATA) proposed by Ares (Ares et al. 2015;Jaeger et al. 2017) are the two main dynamic sensory evaluation methods.The TDS method requires evaluators to select a dominant aroma attribute, which is fast and targeted, and is more conducive to product comparison and development.The TCATA method, originated from CATA, requires the evaluators to screen the set description words in the questionnaire.
Among these sensory evaluation methods, TI, QDA, TDS, and TCATA are the most popular in aroma perception evaluation.The TI method is simple and QDA is detailed, whereas the operation of TDS and TCATA can effectively screen out the significant attribute.During food oral processing, more useful and effective sensory information could be obtained by combination of the TDS/TCATA and QDA methods.

Novel aroma perception methods by brain
Aroma perception arises from the central integration of chemistry, physics, psychology, physiology, neuroscience, cognitive science, and others (Eldeghaidy et al. 2011).There are some noninvasive detection technologies to characterize the aroma perception according to the brain response of consumer including electroencephalography (EEG), functional magnetic resonance imaging (fMRI), electrobulbogram (EBG), magnetoencephalography (MEG), and positron emission tomography (PET) (Anderson et al. 2003;Iravani et al. Figure 6. the olfactory receptor and their corresponding activated aroma compounds (a, the aroma compounds with concentration for 50% of maximal effect (eC50, µM); a' , the name of olfactory receptors; b' , the corresponding activated aroma compounds; data were showed in table s4.B, aroma compounds without eC50 values; a' , the name of olfactory receptors; b' , the corresponding activated aroma compounds; data were showed in table s5).supplement data presented in table s3.
2019).The EEG reflects the postsynaptic potentials of neurons, and these electrical changes are reflected in the EEG recorded at the scalp within milliseconds thus making this methodology outstanding for tracking rapid shifts in brain function.The PET and fMRI can provide detailed images of the human brain that reflect localized changes in cerebral blood-flow and oxygenation induced by motor, sensory, or cognitive tasks.Although they have relatively low temporal resolution, they exhibit a better spatial resolution than EEG and MEG.However, PET and fMRI are not suitable for patients with smell or taste disorders.These methods are very important for food-related emotional response and food preference, providing useful and detailed data to support the understanding of consumer responses to food.Small et al. (2005) used the fMRI technique to elucidate the retronasal olfaction perception resulted from the activation of a network of regions consisting of the orbitofrontal cortex, frontal operculum, ventral insula, amygdala, and anterior cingulated cortex (Figure 5D).The olfactory bulb 3D MRI technique has proven that, using same odorants for orthonasal and retronasal stimulations, the retronasal stimuli required the double orthonasal odor concentration for obtaining similar response amplitudes (Sanganahalli et al. 2020).Retronasal maps were dominant in caudal and lateral regions whereas orthonasal maps were dominant in dorsal-medial regions.Iravani et al. (2019), proposed that the EBG measure method to detect the odor perception signal from human OB was valid and reliable.Based on these electrophysiological and analytical imaging methods, the aroma perception signal can be concretized and quantified to better decode the neural signal mechanisms.These differences in OB encodings are likely to underlie the variances in aroma perception between biologically critical routes for odorants among different people.

Concluding remarks and perspectives
This review systematically summarized and discussed the aroma release from food matrix to nasal cavity during food oral processing, including the influencing factors on aroma release, analytical methods of aroma perception, and the mechanisms of olfactory signal translation.It concludes that the aroma release is affected by many factors, including individual variance and aroma monitoring instruments.The dynamic sensory evaluation method is an useful tool to clarify the aroma perception of psychophysical value.As the concentration factor of odorant, the binding with OR is the key breakthrough.It is necessary to establish a new approach to integrate the relationship among the binding parameters, dose effects, aroma attributes and cross-modal interaction of aroma-taste to obtain a more accurate aroma prediction model for artificial intelligence flavoring.Electrophysiology and imaging technology are powerful methods to link the gap between flavor chemistry and psychological perception.These results can help in the scientific screening and prediction of the favorite products of consumers.
Some challenges and perspectives for the future studies: (1) Interdisciplinary subjects should be combined to clarify the oral parameters effect on the aroma release.(2) Characterization of the key odorants perceived in retronasal aroma perception should be further investigated.(3) The structure decoding of ORs and their aroma activator (EC50 values) should be further investigated for decoding the aroma perception mechanisms.(4) Combination of the psychology and neuroscience techniques to solve the signal analysis of aroma perception during food consumption is an important research direction in future.

Figure 1 .
Figure 1.recent research trends in aroma release and perception during food oral processing (a: Comparative analysis results of publication articles among different countries with the theme of aroma release and perception.retrieving the aroma release and perception at web of science; B: overlay visualization of the title and abstract in selected 300 references from 1980-2021.the result is conducted by vosviewer version 1.6.16,leiden university, the netherlands).

Figure 2 .
Figure 2. summary of the analytical methods for aroma perception affected by the nasal air flow (a, mechanistic mathematical model reference from trelea et al. (2008); B, human olfactory experiment of nasal airflow control reference from Yao et al. (2020); C, determination of the air flow parameters by laser doppler anemometry (lda) reference from Berger et al. (2021); d, computed tomography scanning results of the human nasal cavity and paranasal sinuses reference from tesch, Meyer-szary, Markiet, and skorek (2021); e, 3d model of orthonasal flow reference from ni et al. (2015); F, 3d visualization of the healthy nasal cavity generated by the statistical shape model reference from Brüning et al. (2020).).

Figure 3 .
Figure 3. summary of the analytical methods and the detection results of aroma compounds during food oral processing (a, volatiles detected by different analytical methods; B, statistical the different aroma compounds release and perception during oral processing.).

Figure 5 .
Figure 5. the mechanisms of aroma perception and signal transduction.(Figure verified from the references Briand et al. 2002; small et al. 2005; trimmer et al. 2017; Castro et al. 2021; a, the aroma compounds binding with binding protein; B, the mechanisms of aroma compounds binding with olfactory receptor;C, the olfactory system of olfactory system; d, the process of olfactory perception in cerebral cortex.).