Godlessness in the Land of Gods: Unravelling Atheism in a Hindu Context

1.1 Understanding atheism

Atheism, or the absence of belief in god(s), is pluralistic and diverse and varies substantially by cultural context (Lee 2015; Rahmani et al. 2024). Numerous theories of the decline of theistic beliefs have been proposed, but we will focus on two of them: a lack of religious socialisation and the role of secular education.

One of the key factors shown to lead to religious decline is a lack of religious socialisation (Gervais et al. 2021; Lanman 2012; Turpin & Willard 2022; Willard & Cingl 2017). Though religious socialisation can take many forms, the one most relevant to our purposes is socialisation through CREDs (Henrich 2009). Specifically, CREDs theory suggests that people exposed to a belief are more likely to adopt and maintain that belief if they see those professing the belief acting in ways consistent with that belief. These actions signal to the learner that the beliefs are sincerely held by the practitioners.

In the context of religion, new members of a group (generally children) who are exposed to people investing time, money and other resources in religion are more likely to adopt and maintain religious belief than those less exposed (Langston et al. 2020). Research on this has been predominantly conducted in Europe and North America, but some cross-cultural survey work has been done (Maij et al. 2017). To our knowledge, there has been no examination of the role of CREDs in India.

Formal education has also been shown to play a role in religious decline in the West. Researchers have found that the more exposed one is to secular education, the more likely one is to question the role of religion in secular societies (Mitchell 2017; Schwadel 2011). Secular education exposes students to different lifestyles and ways of being, as well as different cultural norms and role models they can follow (Scheitle & Corcoran 2023).

The broad body of research on atheism has been developed with research on Western, Education, Industrialised, Rich and Democratic (WEIRD; Henrich et al. 2010) samples, and usually following Abrahamic religions. They are assumed to apply similarly in other types of societies and religions, but little evidence on this has been collected. Specifically, very little is known about what atheism looks like among Hindus, the third largest religious group in the world (Hackett & McClendon 2017) and how diverse it might be (Binder 2022; Quack & Binder 2018; Ståhl 2021). This paper aims to demonstrate that understanding the diversity in Hindu populations is important in understanding atheism in this context and explore some of the diverse forms of atheism present within different Hindu groups.

1.2 Understanding Hinduism

To better understand Hindu atheism, it is first important to understand the diversity at the core of Hinduism. Hinduism is the world’s largest polytheistic religion (Hackett & McClendon 2017). It contains thousands of gods, whose worship varies substantially between different linguistic communities and castes (Michaels 2004). Post-colonization, India was reorganised based on the dominant language that was spoken in the area (Mondal 2014) in order to preserve the unique needs and cultural, economic, and religious practices of these linguistic communities (Singh & Dhussa 2020).

Indian linguistic communities can vary substantially, especially in religious practice. Gods dominant in one community may be minor in another. Religious texts considered sacred in one community may be ignored in another. Because of this, Hinduism specialists are often reluctant to make general claims about the religion and instead argue that statements can only be made about specified regional areas (Michaels 2004).

Each linguistic community is further segregated by caste (Badiani et al. In prep.). The caste system is an endogamous hierarchical system that groups people based on levels of perceived purity, which determines communal access, health, and wealth (Johri & Anand 2022). Higher castes are considered purer, and hence have more political and societal power. The opposite is true for lower castes (Claveyrolas 2015; Thorat et al. 2020).

The caste system consists of three distinct identity systems: varna, jati, and governmental caste codifications. Briefly, the varna system divides people based on their vocation (Goghari & Kusi 2023). Priests (Brahmins) sit on the top of the varna system, followed by warriors (Kshatriyas), then tradespeople (Vaishyas), and menial labourers (Shudras) at the bottom. Avarnas fall outside the varna system for performing ‘impure’ tasks such as dealing with human excreta and carcasses (Sankaran et al. 2017).

The jati system refers to an endogamous and closed system (Johnson & Johnson 2024). People marry and reproduce within their jati, and usually all members of the jati work in the same profession. There are thousands of jatis, but most are regionally specific (Shah 2023). This means the jati of washermen in one part of the country is different to the jati of washermen in another part of the country (Channa 1991). Varna and jati have become terms that are now used interchangeably (Chauhan 2018; Sharma 1990). Jatis are now, though not historically, organised into a hierarchical system, such that each jati slots into one of the five rungs of the varna system and are considered sub-castes (Kumar 2014).

The governmental system of caste was developed post-independence to deliver affirmative action. This system divides people into the General category (non-marginalised people), the Other Backward Classes category (OBC; minority castes and religious groups), the Scheduled Caste category (those who are on the lowest rungs, often the avarnas), and Scheduled Tribe category (indigenous people who previously lived nomadic lives) (Sankaran et al. 2017). The scheduled castes and tribes are known as the ‘Dalits’.

Indian-Hindus can refer to their caste in terms of their varna, their jati, or their governmental caste category. For simplicity, this study categorizes castes as follows: ‘high caste’ refers to Brahmins and jatis associated with them; ‘middle caste’ includes those who are not at the top of the caste hierarchy but still benefit from it; and ‘low caste’ refers to marginalized groups who face discrimination because of the caste system (Badiani et al. In prep.).

Religious beliefs vary significantly by caste. People in high castes often have the right to read scriptures and conduct ceremonies. People in middle castes can access all religious spaces, but do not necessarily have the right to perform religious ceremonies. People from the low castes are often not allowed to enter temple spaces due to perceived impurity. Thus, religiosity is highly moderated by caste, a factor often overlooked in the study of Hinduism. Here, we aim to understand how variation in Hinduism between linguistic communities and castes affects theistic beliefs.

1.3 Gujaratis and Maharashtrians ethno-linguistic groups

Gujaratis and Maharashtrians are neighbouring communities situated in the west of India. Until the 1960s, they were one state (‘Bombay state’), but were divided due to linguistic, cultural, and normative differences (Sengupta 2023). The Gujaratis are one of the largest merchant communities in India (Mani 2001), engaging in money-lending, trading, and financing (Bela 2022). The Gujarati community owns a significant portion of the grain, textile, metal, and paper trade in India (Kakodkar 2014) and dominates the global diamond trade (Duttagupta 2003). Despite making up only five percent of the country’s population, they are responsible for one fourth of India’s imports and exports (Modi 2023).

Since trade is considered important in the community, Gujaratis often dismiss the importance of formalised education and focus instead on learning on the job. Businesses are usually passed down to male heirs in each generation. Children are encouraged to understand the inner workings of the business from a young age, even if this comes at the cost of excelling in formal education. Additionally, since enterprise relies on men and the reputation of the family, the community is highly patriarchal(Singh & Sebastian 2018).

Maharashtrians, conversely, rely on salaried employment, especially in urban areas (Anandan 2018; Bela 2022). Their chances of finding employment depend on education and vocational skill (Sawai 2009). Consequently, Maharashtrians prioritise education. The literacy rate within the state is 82.3%, higher than the national average of 73% (Planning Department 2017). Either due to the exposure to and importance placed on secular education, or due employment opportunities being removed from the reputation of the family, the state is also less patriarchal than Gujarat (Badiani 2025). In Maharashtra, education for women is free up until high school (Planning Department 2017). The female labour force participation rate in the state is 31%, higher than the national average (18.6%) and significantly higher than that of Gujarat (16.8%; Pandey 2020; Pandey & Mitra 2020).

Since the vocational habits of the two communities are so different, the caste structure within the communities is also distinct (Badiani et al. In prep.). In Gujarat, traditional caste divisions are preserved. Brahmins are considered the upper-caste, while Shudras and Dalits are still associated with shame. The middle rungs of the caste system are more complicated. Most people identify as a Vaishya or a tradesperson since that is the most reputable profession, and the warrior identity has been absent in Gujarat (Badiani 2025).

In Maharashtra, the Brahmins have preserved their upper-caste status. However, the middle and the low rungs have become highly enmeshed, leading to a highly complicated caste identity. Due to the importance of education, Maharashtra has become a hub of anti-caste resistance (Patil 2024). Many people on the lower rungs of the caste system have joined the ‘Maratha’ movement, which was initially aimed at overthrowing the ‘Mughals’ and British, but later revised to resist the Brahmin supremacy within the state (Deshpande 2004). ‘Maratha’ has now become a caste group comprising both middle and lower rung members. Some people, however, cannot escape the caste system because of their last names. Others, as a form of anti-caste resistance, intentionally choose to continue to associate with the lower rungs.

1.4 The present study

The present study was part of a larger project investigating the cultural evolutionary roots of Hinduism (Badiani 2025). For this larger project, we conducted qualitative interviews (which involved collecting some quantitative data) with both Gujaratis and Maharashtrians. The findings reported in this paper were incidental to the main aims of the interviews, which explored the nature and importance of theism in the two communities. In the interviews, many people spoke about a lack of theism or atheism, which varied by the linguistic community and caste. This paper reports on factors contributing to the decline of theistic belief and the diverse types of atheism observed within the Gujarat and Maharashtrian linguistic communities.

2.1 Research design

Data was collected through in-depth in-person interviews and ethnography. We approached the data with the understanding that belief decline is influenced by both subjective experiences and structured environmental processes (see Al-Saadi 2014). Given the absence of prior theoretical frameworks derived from Hindu contexts, exploring participants’ interpretations aided in assessing existing theories for their applicability to Hindu populations and identifying gaps.

2.2 Participants

Gujarati data was collected from the city of Baroda (N = 36; mean age = 46.6). Maharashtra data collected from the city of Kolhapur (N = 39, mean age = 41.8). Both cities are classified as tier-two cities, and thus comparable in population, infrastructure and development (Ministry of Personnel 2016). The exact participant demography breakdown by caste, and education level completed can be found in Table 1.

All participants from Gujarat were born in Baroda to Hindu parents who themselves were born either in Baroda or in nearby areas within Gujarat. All participants from Maharashtra were born in Kolhapur to Hindu parents either from Kolhapur or nearby areas within Maharashtra. This was to ensure that all participants were socialised into Hinduism within the sensitive period of belief acquisition (Cheung et al. 2011; Lanman 2012). Participants’ personal religious beliefs were considered less important than their parents’, as we wanted to understand the normative influence of Hinduism. This allowed us to capture atheism among the participants.

2.3 Data collection

Data was collected between February and October, 2022. Participants in each site were recruited through initial community connections, which were expanded via snowball sampling. All interviews were conducted in the regional languages (Gujarati or Marathi). Each interview began with demographic questions. We then asked about the participant’s religious beliefs. These questions can be grouped into two categories:

1. The strength of participant’s religious beliefs

2. The frequency with which the participants behaviourally engaged with religious practices and rituals

These questions were originally posed on a five-point scale, but some participants found it easier to answer in terms of a percentage that reflected either the strength of their belief or the frequency with which they engaged in religious behaviours. Some participants chose to not quantify their answer and instead answered the questions qualitatively, giving examples and explanations for their attitudes and behavioural choices. The rest of the interview was a free-flowing discussion, where participants detailed their experience with the religion. All interviews were conducted by the first author with help from a research assistant in the Kolhapur site and were transcribed by native speakers of the languages.

2.4 Data analysis

Our main findings come from the qualitative interview data, which was analysed using reflexive thematic analysis. This approach was chosen because it allows for identifying and interpreting patterns of meaning across participants’ narratives while also acknowledging the researcher’s active role. The first author’s knowledge of the two communities and Hindu culture guided data analysis.

Data analysis occurred in two phases. The first phase involved inductive coding, focusing on participants’ meaning-making. In this stage, statements were grouped by shared meanings to identify initial patterns. The second phase of theme-development was deductive and theory-informed. Here, themes were constructed through relevant theories of atheism and religiosity, allowing the analysis to move beyond descriptive groupings toward abstract and conceptual understanding (Naeem et al. 2023). This hybrid inductive–deductive process aligns with previous research using similar approaches to produce a richer, more evaluative, and reflexive account of meaning-making (Fereday & Muir-Cochrane 2006).

Interviews from Baroda and Kolhapur were analysed separately. The codes were later integrated to identify two semantic themes (‘exposure to CREDs’ and ‘role of education’) and one latent theme (‘caste-based religious variation’). The latent theme was developed by combining three semantic codes: ‘brahmin religious fortification’, ‘fuzzy fidelity among the middle castes’ and ‘absolute rejection among the low castes’ (Figure 1).

Figure 1

Quantitative data from each site was collapsed into a single composite religiosity score, i.e., a continuous measure summarizing overall religiosity across belief and practice. Higher scores indicated higher religiosity. In Kolhapur, belief-related items were recorded as percentages rather than ordinal values to enhance participant ease. These were rescaled into five-point intervals (0–20% = 1, 20–40% = 2, etc.) to align with the five-point scale used in Baroda.

3.1 Qualitative findings

In Gujarat, almost none of the participants reported atheistic beliefs, and many only reluctantly admitted to having lower-than-expected religiosity, possibly due to high exposure to CREDs through long and elaborate daily prayer rituals. These practices are sustained by religious norms and the economic niche of enterprise and trade within tightly knit Gujarati community structures. For Gujaratis, maintaining visible religiosity is both culturally valued and economically advantageous. The relatively lower prioritization of secular education reinforces traditional religious adherence.

In contrast, Maharashtrian participants were more likely to report atheism, which we link to both lower exposure to costly CREDs and a greater emphasis on secular education. This difference, too, is rooted in economic context: Maharashtrians were more likely to engage in salaried or bureaucratic professions where religious signalling holds less economic utility. However, the nature of atheism varied significantly by caste. Brahmins tended to retain their religious beliefs by participating in elaborate rituals and engaging in religious education. Middle-caste Maharashtrians exhibited what we refer to as ‘fuzzy fidelity’, a flexible approach to religiosity in which individuals were behaviourally religious but cognitively less so. Lower-caste Maharashtrians, in contrast, often rejected Hinduism entirely, framing it as a system of oppression, and embraced scientific and rationalist frameworks. These themes are elaborated upon with participant quotes in the sections below.

3.1.1 Exposure to CREDs

For Gujaratis, daily prayer rituals are elaborate, taking up hours of one’s day. High caste Gujaratis generally prayed to Shiva, and these rituals were performed by men. Middle and low caste Gujaratis prayed to Krishna, and rituals were performed mostly by women. For both Shiva and Krishna, participants engaged in ‘Sringar’, literally translated as ‘adornment’. The process of Sringar differed between Gods, but involved dressing and adorning them. Sringar, was accompanied with ‘puja’, which is the chanting of prayers with either lamp, a bell or beaded necklace (like a rosary) during and after the gods are being adorned. The word Sringar is often associated with the term ‘seva’, which means servitude, suggesting one must devote their time, effort and capital while performing Sringar.

For Shiva, Sringar involves adorning the ‘Shiv-ling’. It is a cylindrical shaped stone, which is decorated with flowers, leaves, and milk. Prayer to Shiva is long and elaborate, occurring over several times a day, making it time-intensive. The image on the left in Figure 2 is of a flower adornment of Shiva done by a participant’s father-in-law. The image on the right is of their home Shiva idol, adorned with leaves and oil that is dripping onto the Shiv-linga.

Figure 2

For the middle and low caste Gujaratis in our sample, Krishna was the main God. The baby version of Krishna or ‘baal Krishna’ was primarily worshipped. This involved waking up baby Krishna from his bed in the morning and bathing and clothing him. He was then adorned with ornaments, fed, and given water. Some people put out toys for him. This routine was accompanied with puja and chanting, often lasting a few hours. In the afternoon, Krishna is put to bed for a nap and woken again for evening prayers. At night, Krishna is served dinner and put to bed. Since Sringar for Krishna resembles looking after a child, it is usually conducted by women. The time taken in this process is illustrated here:

‘R: I prepare hot water to bathe Lala. I also prepare tulsi and water for him to drink. Once everything is ready, I sit. Then, I take Lukewarm water and clean handkerchief, give bath to Lala. I have a picture of Shreeji baba, so I use the same napkin to wipe his photo. Then put him (Lala) in new clothes, applying perfume in it… It takes me 2.5 hours daily.’

- r3_Baroda

Figure 3 displays the Sringar of Krishna. On the right is an image of baal Krishna after Sringar has been completed. On the left there is an image of a mosquito-netted bed one participant bought for Krishna to sleep in.

Figure 3

Sringar is a time and resource intensive act performed on a daily basis in almost all Gujarati homes we visited and often observed by children. This includes the first author, who observed her grandmother perform Sringar everyday as she sat next to her. Sringar thus acts as a CRED, potentially bolstering one’s religious dedication.

In Maharashtra, by contrast, prayer rituals were quick and less frequently performed. Most Maharashtrians interviewed were either working or studying, which meant that they did not have the time to engage in elaborate prayer rituals like Gujaratis. Thus, the younger generation in Maharashtra where not as frequently exposed to CREDs as are the Gujarati.

‘I: How much time do you think you dedicate to praying every day?

R: 10 to 15 minutes. Not every day but 2 to 3 times a week’.

- r9_Kolhapur

This minimalistic prayer, lacking evidence of CREDs, is visible in the temples in the Maharashtrian households. Gods are not adorned. Instead, idols, or simply photos, are briefly bowed to before people leave for work. This can be seen in Figure 4. Pictures from the homes of two different participants, with different levels of affluence, show a similar temple structure. This suggests it is not affluence but the approach to religion, determined by the economic structure of the community that influences the simplicity of temples and prayer rituals. The temples in Figure 4 demonstrate much lower level of CREDs among Maharashtrians than Gujaratis.

Figure 4

3.2 Quantitative findings

The composite religiosity score, derived from three items assessing religious support, comfort, and influence, was higher in Baroda (M = 4.44, SD = 0.68, range = 3–5) than in Kolhapur (M = 3.25, SD = 1.54, range = 1–5). A Welch’s t-test confirmed a significant difference, t(53.09) = 4.41, p < .001, 95% CI [0.65, 1.73], with a large effect size (d = 0.99). Participants from Baroda thus showed markedly higher religiosity than those from Kolhapur. Frequency of prayer was also higher in Baroda (M = 4.65, SD = 0.80) than in Kolhapur (M = 3.87, SD = 1.56), consistent with qualitative accounts suggesting greater exposure to CREDs in Baroda.

4.1 Caste-determinant pathways to atheism

Our findings suggest that building and testing theories of atheism among Hindus cannot ignore the role of caste. The lack of belief observed among middle-caste Maharashtrian Hindus is less a critical and conscious rejection of religion and more a product of passive secularisation. This secularisation occurs gradually and non-consciously as societal changes reduce the need for religious engagement. The role of societal development in fostering atheistic thought is well documented. As secular institutions and income create a more secure world, people’s need for the security offered by religion can decline (Norris & Inglehart 2004; Storm 2017). Urbanisation and industrialisation can further reduce the adaptive value of religion, which lies in part in its ability to bolster cooperation through supernatural monitoring and punishment (Norenzayan 2013). In urban and industrialised spaces, technology and legislative systems take over this function, making religious enforcement redundant.

Within Hinduism, this pathway is afforded to middle caste people who have the privilege to ‘opt out’ of religiosity while still benefiting from the cultural and social structures upheld by religion. They participate in group-level festivals and rituals instead of engaging in conscious personal devotion. In doing so, they end up upholding the relevance of religious structures in the cultural nexus without deep cognitive engagement.

While we largely attribute the weaker religious commitment of middle-caste Maharashtrian Hindus to lower exposure to CREDs, this alone does not fully account for these differences. We propose that religion plays an additional role for Gujaratis, helping them signal group membership and foster trust among strangers, which is crucial for enterprising trade (Badiani, 2025; Power, 2017). Thus, we hypothesise that participation in time-consuming and costly rituals has benefits for this community beyond the transmission of belief to new generations.

Among lower caste individuals, maintaining even a performative relationship with religion implies participating in their own subjugation. As a result, reflective atheism becomes more common, but only when lower exposure to CREDs is coupled with access to secular education and anti-caste literature. This combination creates an intellectual foundation for the critical rejection of Hinduism. Atheism among lower-caste Hindus functions as active resistance, not a process of religion slowly fading away through decreased relevance. This atheism is especially evident in Maharashtra, as it is the home of notable anti-caste revolutionaries such as Dr. Ambedkar, Jyotiba Phule, Savitribai Phule, and Narendra Dabholkar, who have promoted anti-caste dissent in the form of reflective atheism.

In practice, caste is much more complex than simple categories of high, middle, and low, with thousands caste stratifications. Each caste strata deals with its own experiences of subjugating and being subjugated by other castes. This is likely to play a role in one’s willingness to reject religion.

We also do not consider class here. Class can sometimes garb one’s caste positioning, allowing some in lower castes the privilege to maintain fuzzy-fidelity without necessitating active resistance. Future research into the theistic expression of each caste within a single community would greatly further our understanding. While cross-cultural research has its merits, it can overlook unique features of each community, hindering our understanding of the full range of atheistic/theistic expression.

Our exploration of samples from two linguistic groups in two Indian cities illustrates a need to look at the diversity of different Hindu communities and their approach to religion, but our sample vastly underrepresents this diversity and should be expanded in future research. Even when such comparisons are not possible, work on atheism and religion in a Hindu context should take into account the specific linguistic group and their approach to religion and irreligion.