Sensitivity analyses were performed by excluding one study from the meta-analysis and performing more than two comparisons and multiple analyses to test the stability of the results. A study was considered to have influenced the results if the overall correlation changed by more than 10% after excluding it, or if the significance, direction, or difference in outcomes changed. We examined the symmetry of the funnel plot and applied the Egger test to assess evidence of publication bias [13]. Where heterogeneity was high, we excluded studies assessed as being at high risk of bias in repeated pooled analyses [14].

Upon retrieval, we included a total of 21 RCT studies (Tables 3, 4). Following assessment by Cochrane RoB 2, two studies [15, 16] were identified with some issues overall within this review. This was attributed to biases in participant and personnel blinding. Among the remaining 19 studies, 11 were found to have potential high risk of bias, with 8 studies [17,18,19,20,21,22,23,24] having potential risks due to other biases, 3 studies [25,26,27] with potential risks in random sequence generation, and 8 studies [28,29,30,31,32,33,34,35] assessed as low risk (Figs. 2, 3).

The majority of trials (15) were conducted in Asia, including China [23, 35], Japan [22, 24, 25, 28, 31, 33], Korea [19], Thailand [21], Indonesia [26], and Iran [18]; followed by the United States (1) [34], Australia (1) [27], Poland (1) [15], the Netherlands (1) [32], Egypt (1) [16], and Turkey (1) [17]. All 20 studies involved both men and women, with only 1 study targeting only healthy young women [15]. 12 studies [18, 19, 22,23,24,25, 28, 30, 31, 33,34,35] were conducted only in middle-aged and older adults ( ≥ 50 years), 5 in adults aged 18-60 years [15, 20, 21, 29, 32], 3 in infants and children (1-16 years) [16, 26, 27], and 1 study did not specify the age of the participants [17], only that it was conducted in preterm infants. 5 trials of the study were in older populations with memory impairment including Alzheimer's, mild cognitive impairment, and memory loss [18, 24, 25, 31, 33], 2 trials were in adults with depression or obesity [21, 32], 2 trials were in preterm infants [17, 27], and 1 trial was in children with ADHD [16]. One trial was conducted in elderly patients with epilepsy [35], and the rest were conducted in healthy populations, with follow-up periods ranging from 8 weeks to 3 years.

Twenty-one papers included baseline cognitive function measures. Meta-analyses summarised cognitive outcomes from 21 research trials; of these, 12 reported outcome measures of overall cognitive function, 7 reported outcome measures of processing speed, 14 reported outcome measures of executive function, 18 reported outcome measures of delayed memory, 14 reported outcome measures of attention, 11 reported outcome measures of spatial function, and 11 reported outcome measures of spatial function. Outcome measures of spatial functioning were reported in 14 trials, and outcome measures of verbal ability were reported in 15 trials.

Twelve studies involving 1688 subjects (999 infants and children, 103 prime-aged, and 586 middle-aged and older adults) assessed the effects of probiotics on overall cognition. Two of the studies used the MMSE, four used the RBANS, two used the Bayley-III, and the remaining four used the Computerised CogState Test, NIH Toolbox, Cognitrax Test, and the BSID-II, respectively. Figure 4 shows the results of the summary analyses of the cognitive effects of probiotics on different age groups (Fig. 4). Pooled analysis results. We used a random effects model and probiotics did not significantly promote cognitive development in infancy and childhood compared to controls (SMD = 0.11, 95% CI [-0.02, 0.23], P = 0.09). Similarly, in prime adulthood, probiotics did not significantly affect cognitive performance (SMD = -0.06, 95% CI [-0.44, 0.33], P = 0.77). In middle and old age, probiotics demonstrated a positive effect in maintaining cognitive function (SMD = 0.56, 95% CI [0.19, 0.93], P = 0.003). However inter-study heterogeneity was high, and after excluding trials [18, 23, 25], a sensitivity meta-analysis showed that heterogeneity was reduced to (I = 37%) and did not impact the results (SMD = 0.40, 95% CI [0.13, 0.67], P = 0.004). Visual inspection of the funnel plots did not reveal any publication bias (Fig. 5).

A total of 14 studies were conducted involving 1008 participants, including 229 infants, 377 adolescents, and 402 middle-aged and older adults. These studies reported 12 tests including Stroop Interference Test, RBANS, WCST, ADAS-Jcog, Computerised CogState Test, ADL, NIH Toolbox, Cognitrax Test, TMT-B, Bayley-III, APT, and MoCA.Pooled analyses showed that probiotics had a significant positive effect on executive functioning in infancy and early childhood (SMD = 0.30, 95% CI [0.04,0.56], p = 0.03). However, probiotics had no effect on executive function in young adults (SMD = 0.07, 95% CI [-0.25, 0.40], P = 0.65) as well as middle-aged and older adults (SMD = 0.27, 95% CI [-0.04, 0.58], P = 0.09). These results are shown in Fig. 6.

A total of 369 individuals were involved in seven research trials, consisting of 62 infants and children and 307 middle-aged and older adults. These trials examined the impact of probiotics on processing speed using respected cognitive assessment tools such as the WAIS-IV, RVIP, NIH Toolbox, Cognitrax Test, RBANS, TMT-A, or WPPSI. The combined analysis revealed a significant improvement in processing speed among middle-aged and older adults who consumed probiotics compared to the control group (SMD = 0.37, 95% CI [0.11, 0.63], P = 0.006). However, the effect of probiotics on processing speed in infants and children was not statistically significant (SMD = 0.20, 95% CI [-0.31, 0.71], P = 0.44) (Fig. 7).

In 16 RCTs involving 1156 subjects, 584 subjects received the intervention and 572 served as controls. Sixty-two infants and children, 377 young adults, and 717 middle-aged and older adults were included. These trials reported the effects of 13 tests such as the Computerised CogState Test, WAIS-IV, SRT, RBANS, ADAS-Jcog, MMSE, NIH Toolbox, WPPSI, RAVLT, MoCA, CVLT, and CERAD-K on testing delayed memory. Pooled analysis of the 16 tests showed no significant effect of probiotics on memory function in infants and children as well as adolescents (SMD = 0.20, 95% CI [-0.31, 0.71], P = 0.44) (SMD = 0.03, 95% CI [-0.18, 0.24], P = 0.77) (Fig. 8). However there was a statistically significant difference between the intervention and control groups in the middle-aged as well as the elderly (SMD = 0.51, 95% CI [0.25, 0.78], P = 0.0002). It is worth noting that heterogeneity between groups was high (I = 67%). We performed a sensitivity analysis and after excluding articles [18], heterogeneity was reduced to (I = 9%), but this did not affect the overall findings (SMD = 0.37, 95% CI [0.21, 0.54], P < 0.00001).

In 13 RCTs involving 1049 subjects, 522 subjects received the intervention and 527 served as controls. Seventy-six infants and children, 265 young adults, and 708 middle-aged and older adults were included. These trials reported the effects of the Computerised CogState Test, DSF, RBANS, CPT, ADAS-Jcog, Cognitrax Test, MMSE, APT, MoCA, and CERAD-K 10 tests on the testing of attention. Aggregate analysis of these 13 tests revealed that probiotics did not show significant improvement in attention across different age groups, including infants and children, adolescents, and middle-aged and older adults (SMD = 0.27, 95% CI [-0.19, 0.72], p = 0.25; SMD = -0.05, 95% CI [-0.38, 0.29], p = 0.79; SMD = 0.13, 95% CI [-0.13, 0.39], P = 0.32)(Fig. 9).

A total of 10 RCTs recruited 729 participants, of whom 370 were allocated to the intervention group and 359 to the control group. Of these, 62 were infants and the rest were middle-aged and older adults. These trials investigated the visuospatial effects of eight different tests, including the WAIS-IV, RBANS, ADAS-Jcog, Cognitrax Test, MMSE, WPPSI, MoCA, and CERAD-K. Pooled analyses of these tests showed that the visuospatial effects of probiotics on infants and young children did not differ significantly between the intervention and control groups (SMD = 0.20, 95% CI [-0.31, 0.71], p = 0.44). However, in the middle-aged and elderly group, probiotics were found to significantly improve visuospatial abilities (SMD = 0.35, 95% CI [0.09, 0.62], p = 0.008) (Fig. 10). Although there was some heterogeneity (I = 65%), sensitivity analyses showed that excluding two articles reduced heterogeneity to 47% and had no effect on the results (SMD = 0.35, 95% CI [0.09, 0.62], p = 0.008).

A total of 14 RCTs recruited 1546 participants, of whom 776 were assigned to the intervention group and 770 to the control group. 723 were infants, 200 young adults, and 623 middle-aged and older adults. These trials investigated the language proficiency effects of 11 different tests including the Computerised CogState Test, WAIS-IV, RBANS, VLT, ADAS-Jcog, CFT, Bayley-III, CVLT, FAS, MoCA, and CERAD-K. Pooled analyses showed that the effects of probiotics on the speech and language abilities of infants, middle-aged and older adults had no significant effect on language skills between the intervention and control groups (SMD = -0.02, 95% CI [-0.16, 0.13], p = 0.84), (SMD = 0.10, 95% CI [-0.11, 0.32], p = 0.34). However, in young people, probiotic supplementation was associated with significant improvements in speech performance as assessed by the Computerised CogState Test, CVLT, and FAS. (SMD = 0.43, 95% CI [0.15, 0.71], p = 0.003) (Fig. 11).